diff --git "a/checkpoint-1280/trainer_state.json" "b/checkpoint-1280/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1280/trainer_state.json"
@@ -0,0 +1,9066 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.20122226807365048,
+  "eval_steps": 160,
+  "global_step": 1280,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00015720489693253945,
+      "grad_norm": 1.3751904964447021,
+      "learning_rate": 0.0,
+      "loss": 3.5741,
+      "step": 1
+    },
+    {
+      "epoch": 0.00015720489693253945,
+      "eval_loss": 3.4173049926757812,
+      "eval_runtime": 2315.7248,
+      "eval_samples_per_second": 3.998,
+      "eval_steps_per_second": 1.999,
+      "step": 1
+    },
+    {
+      "epoch": 0.0003144097938650789,
+      "grad_norm": 1.231239676475525,
+      "learning_rate": 5e-06,
+      "loss": 3.3021,
+      "step": 2
+    },
+    {
+      "epoch": 0.00047161469079761836,
+      "grad_norm": 1.3657807111740112,
+      "learning_rate": 1e-05,
+      "loss": 3.6333,
+      "step": 3
+    },
+    {
+      "epoch": 0.0006288195877301578,
+      "grad_norm": 1.3117496967315674,
+      "learning_rate": 1.5e-05,
+      "loss": 3.3731,
+      "step": 4
+    },
+    {
+      "epoch": 0.0007860244846626972,
+      "grad_norm": 1.4118576049804688,
+      "learning_rate": 2e-05,
+      "loss": 3.612,
+      "step": 5
+    },
+    {
+      "epoch": 0.0009432293815952367,
+      "grad_norm": 1.3155895471572876,
+      "learning_rate": 2.5e-05,
+      "loss": 3.3296,
+      "step": 6
+    },
+    {
+      "epoch": 0.001100434278527776,
+      "grad_norm": 1.2847192287445068,
+      "learning_rate": 3e-05,
+      "loss": 3.2168,
+      "step": 7
+    },
+    {
+      "epoch": 0.0012576391754603156,
+      "grad_norm": 1.1421078443527222,
+      "learning_rate": 3.5e-05,
+      "loss": 3.085,
+      "step": 8
+    },
+    {
+      "epoch": 0.0014148440723928551,
+      "grad_norm": 0.9923035502433777,
+      "learning_rate": 4e-05,
+      "loss": 3.0472,
+      "step": 9
+    },
+    {
+      "epoch": 0.0015720489693253944,
+      "grad_norm": 0.795043408870697,
+      "learning_rate": 4.5e-05,
+      "loss": 2.6666,
+      "step": 10
+    },
+    {
+      "epoch": 0.001729253866257934,
+      "grad_norm": 0.5987974405288696,
+      "learning_rate": 5e-05,
+      "loss": 2.473,
+      "step": 11
+    },
+    {
+      "epoch": 0.0018864587631904734,
+      "grad_norm": 0.4488905668258667,
+      "learning_rate": 4.9999999236547564e-05,
+      "loss": 2.3731,
+      "step": 12
+    },
+    {
+      "epoch": 0.002043663660123013,
+      "grad_norm": 0.3517301380634308,
+      "learning_rate": 4.999999694619029e-05,
+      "loss": 2.2158,
+      "step": 13
+    },
+    {
+      "epoch": 0.002200868557055552,
+      "grad_norm": 0.3045121431350708,
+      "learning_rate": 4.999999312892831e-05,
+      "loss": 2.3351,
+      "step": 14
+    },
+    {
+      "epoch": 0.002358073453988092,
+      "grad_norm": 0.24488244950771332,
+      "learning_rate": 4.9999987784761884e-05,
+      "loss": 2.2693,
+      "step": 15
+    },
+    {
+      "epoch": 0.0025152783509206312,
+      "grad_norm": 0.22892728447914124,
+      "learning_rate": 4.999998091369132e-05,
+      "loss": 2.1006,
+      "step": 16
+    },
+    {
+      "epoch": 0.0026724832478531705,
+      "grad_norm": 0.23219206929206848,
+      "learning_rate": 4.999997251571704e-05,
+      "loss": 2.215,
+      "step": 17
+    },
+    {
+      "epoch": 0.0028296881447857102,
+      "grad_norm": 0.24427154660224915,
+      "learning_rate": 4.999996259083956e-05,
+      "loss": 2.1708,
+      "step": 18
+    },
+    {
+      "epoch": 0.0029868930417182495,
+      "grad_norm": 0.2640205919742584,
+      "learning_rate": 4.999995113905947e-05,
+      "loss": 2.1709,
+      "step": 19
+    },
+    {
+      "epoch": 0.003144097938650789,
+      "grad_norm": 0.26644033193588257,
+      "learning_rate": 4.999993816037749e-05,
+      "loss": 2.1733,
+      "step": 20
+    },
+    {
+      "epoch": 0.0033013028355833285,
+      "grad_norm": 0.2621535062789917,
+      "learning_rate": 4.9999923654794414e-05,
+      "loss": 2.0059,
+      "step": 21
+    },
+    {
+      "epoch": 0.003458507732515868,
+      "grad_norm": 0.2586187422275543,
+      "learning_rate": 4.999990762231111e-05,
+      "loss": 2.0336,
+      "step": 22
+    },
+    {
+      "epoch": 0.003615712629448407,
+      "grad_norm": 0.26732271909713745,
+      "learning_rate": 4.9999890062928566e-05,
+      "loss": 2.0566,
+      "step": 23
+    },
+    {
+      "epoch": 0.003772917526380947,
+      "grad_norm": 0.2357867807149887,
+      "learning_rate": 4.999987097664787e-05,
+      "loss": 1.9529,
+      "step": 24
+    },
+    {
+      "epoch": 0.003930122423313486,
+      "grad_norm": 0.2297009825706482,
+      "learning_rate": 4.999985036347016e-05,
+      "loss": 2.0369,
+      "step": 25
+    },
+    {
+      "epoch": 0.004087327320246026,
+      "grad_norm": 0.20529747009277344,
+      "learning_rate": 4.9999828223396705e-05,
+      "loss": 1.9781,
+      "step": 26
+    },
+    {
+      "epoch": 0.004244532217178565,
+      "grad_norm": 0.18342873454093933,
+      "learning_rate": 4.999980455642887e-05,
+      "loss": 1.9986,
+      "step": 27
+    },
+    {
+      "epoch": 0.004401737114111104,
+      "grad_norm": 0.16487397253513336,
+      "learning_rate": 4.999977936256809e-05,
+      "loss": 1.9063,
+      "step": 28
+    },
+    {
+      "epoch": 0.004558942011043644,
+      "grad_norm": 0.1762266606092453,
+      "learning_rate": 4.99997526418159e-05,
+      "loss": 1.9517,
+      "step": 29
+    },
+    {
+      "epoch": 0.004716146907976184,
+      "grad_norm": 0.16371938586235046,
+      "learning_rate": 4.999972439417394e-05,
+      "loss": 1.7734,
+      "step": 30
+    },
+    {
+      "epoch": 0.004873351804908723,
+      "grad_norm": 0.17309769988059998,
+      "learning_rate": 4.999969461964392e-05,
+      "loss": 1.8732,
+      "step": 31
+    },
+    {
+      "epoch": 0.0050305567018412625,
+      "grad_norm": 0.15772338211536407,
+      "learning_rate": 4.9999663318227683e-05,
+      "loss": 1.7537,
+      "step": 32
+    },
+    {
+      "epoch": 0.005187761598773802,
+      "grad_norm": 0.17521986365318298,
+      "learning_rate": 4.9999630489927126e-05,
+      "loss": 2.0077,
+      "step": 33
+    },
+    {
+      "epoch": 0.005344966495706341,
+      "grad_norm": 0.15462292730808258,
+      "learning_rate": 4.999959613474425e-05,
+      "loss": 1.8576,
+      "step": 34
+    },
+    {
+      "epoch": 0.005502171392638881,
+      "grad_norm": 0.15280336141586304,
+      "learning_rate": 4.999956025268117e-05,
+      "loss": 1.862,
+      "step": 35
+    },
+    {
+      "epoch": 0.0056593762895714205,
+      "grad_norm": 0.14518432319164276,
+      "learning_rate": 4.999952284374006e-05,
+      "loss": 1.8893,
+      "step": 36
+    },
+    {
+      "epoch": 0.005816581186503959,
+      "grad_norm": 0.16087624430656433,
+      "learning_rate": 4.999948390792321e-05,
+      "loss": 1.8658,
+      "step": 37
+    },
+    {
+      "epoch": 0.005973786083436499,
+      "grad_norm": 0.17504698038101196,
+      "learning_rate": 4.999944344523301e-05,
+      "loss": 1.7647,
+      "step": 38
+    },
+    {
+      "epoch": 0.006130990980369039,
+      "grad_norm": 0.17786233127117157,
+      "learning_rate": 4.999940145567191e-05,
+      "loss": 1.8133,
+      "step": 39
+    },
+    {
+      "epoch": 0.006288195877301578,
+      "grad_norm": 0.1628972887992859,
+      "learning_rate": 4.999935793924249e-05,
+      "loss": 1.7731,
+      "step": 40
+    },
+    {
+      "epoch": 0.006445400774234117,
+      "grad_norm": 0.13461466133594513,
+      "learning_rate": 4.9999312895947406e-05,
+      "loss": 1.7558,
+      "step": 41
+    },
+    {
+      "epoch": 0.006602605671166657,
+      "grad_norm": 0.12960125505924225,
+      "learning_rate": 4.99992663257894e-05,
+      "loss": 1.7639,
+      "step": 42
+    },
+    {
+      "epoch": 0.006759810568099196,
+      "grad_norm": 0.10991287231445312,
+      "learning_rate": 4.9999218228771324e-05,
+      "loss": 1.7538,
+      "step": 43
+    },
+    {
+      "epoch": 0.006917015465031736,
+      "grad_norm": 0.11583230644464493,
+      "learning_rate": 4.999916860489612e-05,
+      "loss": 1.715,
+      "step": 44
+    },
+    {
+      "epoch": 0.007074220361964275,
+      "grad_norm": 0.10344280302524567,
+      "learning_rate": 4.999911745416681e-05,
+      "loss": 1.6907,
+      "step": 45
+    },
+    {
+      "epoch": 0.007231425258896814,
+      "grad_norm": 0.10546118766069412,
+      "learning_rate": 4.999906477658651e-05,
+      "loss": 1.7294,
+      "step": 46
+    },
+    {
+      "epoch": 0.007388630155829354,
+      "grad_norm": 0.11775675415992737,
+      "learning_rate": 4.9999010572158465e-05,
+      "loss": 1.7146,
+      "step": 47
+    },
+    {
+      "epoch": 0.007545835052761894,
+      "grad_norm": 0.11109112203121185,
+      "learning_rate": 4.999895484088596e-05,
+      "loss": 1.6939,
+      "step": 48
+    },
+    {
+      "epoch": 0.007703039949694433,
+      "grad_norm": 0.1116517186164856,
+      "learning_rate": 4.999889758277242e-05,
+      "loss": 1.7271,
+      "step": 49
+    },
+    {
+      "epoch": 0.007860244846626972,
+      "grad_norm": 0.11245547980070114,
+      "learning_rate": 4.999883879782132e-05,
+      "loss": 1.7333,
+      "step": 50
+    },
+    {
+      "epoch": 0.008017449743559512,
+      "grad_norm": 0.1150551363825798,
+      "learning_rate": 4.999877848603626e-05,
+      "loss": 1.7036,
+      "step": 51
+    },
+    {
+      "epoch": 0.008174654640492052,
+      "grad_norm": 0.10856381803750992,
+      "learning_rate": 4.999871664742093e-05,
+      "loss": 1.7493,
+      "step": 52
+    },
+    {
+      "epoch": 0.008331859537424591,
+      "grad_norm": 0.10760089010000229,
+      "learning_rate": 4.9998653281979095e-05,
+      "loss": 1.6292,
+      "step": 53
+    },
+    {
+      "epoch": 0.00848906443435713,
+      "grad_norm": 0.0932115837931633,
+      "learning_rate": 4.9998588389714634e-05,
+      "loss": 1.6608,
+      "step": 54
+    },
+    {
+      "epoch": 0.00864626933128967,
+      "grad_norm": 0.09837482124567032,
+      "learning_rate": 4.9998521970631504e-05,
+      "loss": 1.7834,
+      "step": 55
+    },
+    {
+      "epoch": 0.008803474228222209,
+      "grad_norm": 0.08872833847999573,
+      "learning_rate": 4.9998454024733775e-05,
+      "loss": 1.6484,
+      "step": 56
+    },
+    {
+      "epoch": 0.008960679125154749,
+      "grad_norm": 0.08829163759946823,
+      "learning_rate": 4.9998384552025577e-05,
+      "loss": 1.5913,
+      "step": 57
+    },
+    {
+      "epoch": 0.009117884022087288,
+      "grad_norm": 0.09087682515382767,
+      "learning_rate": 4.999831355251117e-05,
+      "loss": 1.6809,
+      "step": 58
+    },
+    {
+      "epoch": 0.009275088919019828,
+      "grad_norm": 0.08675853163003922,
+      "learning_rate": 4.9998241026194884e-05,
+      "loss": 1.6519,
+      "step": 59
+    },
+    {
+      "epoch": 0.009432293815952368,
+      "grad_norm": 0.08463481813669205,
+      "learning_rate": 4.999816697308114e-05,
+      "loss": 1.6234,
+      "step": 60
+    },
+    {
+      "epoch": 0.009589498712884906,
+      "grad_norm": 0.08403950184583664,
+      "learning_rate": 4.999809139317448e-05,
+      "loss": 1.6533,
+      "step": 61
+    },
+    {
+      "epoch": 0.009746703609817445,
+      "grad_norm": 0.08155622333288193,
+      "learning_rate": 4.99980142864795e-05,
+      "loss": 1.6726,
+      "step": 62
+    },
+    {
+      "epoch": 0.009903908506749985,
+      "grad_norm": 0.08056480437517166,
+      "learning_rate": 4.999793565300093e-05,
+      "loss": 1.5881,
+      "step": 63
+    },
+    {
+      "epoch": 0.010061113403682525,
+      "grad_norm": 0.07879023998975754,
+      "learning_rate": 4.999785549274355e-05,
+      "loss": 1.5568,
+      "step": 64
+    },
+    {
+      "epoch": 0.010218318300615065,
+      "grad_norm": 0.07828455418348312,
+      "learning_rate": 4.9997773805712265e-05,
+      "loss": 1.6464,
+      "step": 65
+    },
+    {
+      "epoch": 0.010375523197547604,
+      "grad_norm": 0.08054805546998978,
+      "learning_rate": 4.9997690591912075e-05,
+      "loss": 1.6213,
+      "step": 66
+    },
+    {
+      "epoch": 0.010532728094480142,
+      "grad_norm": 0.07610727101564407,
+      "learning_rate": 4.999760585134805e-05,
+      "loss": 1.5729,
+      "step": 67
+    },
+    {
+      "epoch": 0.010689932991412682,
+      "grad_norm": 0.07693428546190262,
+      "learning_rate": 4.999751958402537e-05,
+      "loss": 1.5444,
+      "step": 68
+    },
+    {
+      "epoch": 0.010847137888345222,
+      "grad_norm": 0.0810319185256958,
+      "learning_rate": 4.99974317899493e-05,
+      "loss": 1.7045,
+      "step": 69
+    },
+    {
+      "epoch": 0.011004342785277762,
+      "grad_norm": 0.07729896157979965,
+      "learning_rate": 4.9997342469125205e-05,
+      "loss": 1.6268,
+      "step": 70
+    },
+    {
+      "epoch": 0.011161547682210301,
+      "grad_norm": 0.07730107754468918,
+      "learning_rate": 4.999725162155855e-05,
+      "loss": 1.658,
+      "step": 71
+    },
+    {
+      "epoch": 0.011318752579142841,
+      "grad_norm": 0.08072328567504883,
+      "learning_rate": 4.9997159247254864e-05,
+      "loss": 1.5045,
+      "step": 72
+    },
+    {
+      "epoch": 0.011475957476075379,
+      "grad_norm": 0.08120577782392502,
+      "learning_rate": 4.9997065346219805e-05,
+      "loss": 1.568,
+      "step": 73
+    },
+    {
+      "epoch": 0.011633162373007919,
+      "grad_norm": 0.08131498098373413,
+      "learning_rate": 4.99969699184591e-05,
+      "loss": 1.6035,
+      "step": 74
+    },
+    {
+      "epoch": 0.011790367269940458,
+      "grad_norm": 0.08395873010158539,
+      "learning_rate": 4.9996872963978584e-05,
+      "loss": 1.5844,
+      "step": 75
+    },
+    {
+      "epoch": 0.011947572166872998,
+      "grad_norm": 0.08502068370580673,
+      "learning_rate": 4.999677448278417e-05,
+      "loss": 1.6661,
+      "step": 76
+    },
+    {
+      "epoch": 0.012104777063805538,
+      "grad_norm": 0.08467952907085419,
+      "learning_rate": 4.999667447488188e-05,
+      "loss": 1.5537,
+      "step": 77
+    },
+    {
+      "epoch": 0.012261981960738078,
+      "grad_norm": 0.19682182371616364,
+      "learning_rate": 4.999657294027782e-05,
+      "loss": 1.5051,
+      "step": 78
+    },
+    {
+      "epoch": 0.012419186857670617,
+      "grad_norm": 0.08586428314447403,
+      "learning_rate": 4.999646987897818e-05,
+      "loss": 1.565,
+      "step": 79
+    },
+    {
+      "epoch": 0.012576391754603155,
+      "grad_norm": 0.08156823366880417,
+      "learning_rate": 4.999636529098928e-05,
+      "loss": 1.6627,
+      "step": 80
+    },
+    {
+      "epoch": 0.012733596651535695,
+      "grad_norm": 0.08715341240167618,
+      "learning_rate": 4.9996259176317486e-05,
+      "loss": 1.5862,
+      "step": 81
+    },
+    {
+      "epoch": 0.012890801548468235,
+      "grad_norm": 0.09664586186408997,
+      "learning_rate": 4.999615153496928e-05,
+      "loss": 1.5741,
+      "step": 82
+    },
+    {
+      "epoch": 0.013048006445400774,
+      "grad_norm": 0.08438891172409058,
+      "learning_rate": 4.999604236695125e-05,
+      "loss": 1.5933,
+      "step": 83
+    },
+    {
+      "epoch": 0.013205211342333314,
+      "grad_norm": 0.08333732932806015,
+      "learning_rate": 4.999593167227006e-05,
+      "loss": 1.5904,
+      "step": 84
+    },
+    {
+      "epoch": 0.013362416239265854,
+      "grad_norm": 0.07945791631937027,
+      "learning_rate": 4.9995819450932455e-05,
+      "loss": 1.5763,
+      "step": 85
+    },
+    {
+      "epoch": 0.013519621136198392,
+      "grad_norm": 0.07682961225509644,
+      "learning_rate": 4.9995705702945304e-05,
+      "loss": 1.5197,
+      "step": 86
+    },
+    {
+      "epoch": 0.013676826033130932,
+      "grad_norm": 0.07547677308320999,
+      "learning_rate": 4.999559042831555e-05,
+      "loss": 1.6825,
+      "step": 87
+    },
+    {
+      "epoch": 0.013834030930063471,
+      "grad_norm": 0.07293456047773361,
+      "learning_rate": 4.999547362705025e-05,
+      "loss": 1.5466,
+      "step": 88
+    },
+    {
+      "epoch": 0.013991235826996011,
+      "grad_norm": 0.07730914652347565,
+      "learning_rate": 4.999535529915651e-05,
+      "loss": 1.5775,
+      "step": 89
+    },
+    {
+      "epoch": 0.01414844072392855,
+      "grad_norm": 0.07689664512872696,
+      "learning_rate": 4.9995235444641565e-05,
+      "loss": 1.5881,
+      "step": 90
+    },
+    {
+      "epoch": 0.01430564562086109,
+      "grad_norm": 0.07754997909069061,
+      "learning_rate": 4.999511406351275e-05,
+      "loss": 1.5037,
+      "step": 91
+    },
+    {
+      "epoch": 0.014462850517793628,
+      "grad_norm": 0.07229866087436676,
+      "learning_rate": 4.999499115577746e-05,
+      "loss": 1.5077,
+      "step": 92
+    },
+    {
+      "epoch": 0.014620055414726168,
+      "grad_norm": 0.07491567730903625,
+      "learning_rate": 4.9994866721443215e-05,
+      "loss": 1.5461,
+      "step": 93
+    },
+    {
+      "epoch": 0.014777260311658708,
+      "grad_norm": 0.07258685678243637,
+      "learning_rate": 4.9994740760517605e-05,
+      "loss": 1.5516,
+      "step": 94
+    },
+    {
+      "epoch": 0.014934465208591248,
+      "grad_norm": 0.07643327116966248,
+      "learning_rate": 4.9994613273008334e-05,
+      "loss": 1.6223,
+      "step": 95
+    },
+    {
+      "epoch": 0.015091670105523787,
+      "grad_norm": 0.0740588903427124,
+      "learning_rate": 4.999448425892318e-05,
+      "loss": 1.5322,
+      "step": 96
+    },
+    {
+      "epoch": 0.015248875002456327,
+      "grad_norm": 0.44172239303588867,
+      "learning_rate": 4.999435371827003e-05,
+      "loss": 1.5498,
+      "step": 97
+    },
+    {
+      "epoch": 0.015406079899388867,
+      "grad_norm": 0.0756363570690155,
+      "learning_rate": 4.999422165105684e-05,
+      "loss": 1.559,
+      "step": 98
+    },
+    {
+      "epoch": 0.015563284796321405,
+      "grad_norm": 0.07251248508691788,
+      "learning_rate": 4.99940880572917e-05,
+      "loss": 1.5903,
+      "step": 99
+    },
+    {
+      "epoch": 0.015720489693253945,
+      "grad_norm": 0.06931837648153305,
+      "learning_rate": 4.999395293698275e-05,
+      "loss": 1.4849,
+      "step": 100
+    },
+    {
+      "epoch": 0.015877694590186484,
+      "grad_norm": 0.07403590530157089,
+      "learning_rate": 4.9993816290138254e-05,
+      "loss": 1.5191,
+      "step": 101
+    },
+    {
+      "epoch": 0.016034899487119024,
+      "grad_norm": 0.07027724385261536,
+      "learning_rate": 4.999367811676655e-05,
+      "loss": 1.5655,
+      "step": 102
+    },
+    {
+      "epoch": 0.016192104384051564,
+      "grad_norm": 0.07320379465818405,
+      "learning_rate": 4.9993538416876093e-05,
+      "loss": 1.4869,
+      "step": 103
+    },
+    {
+      "epoch": 0.016349309280984103,
+      "grad_norm": 0.0726180374622345,
+      "learning_rate": 4.9993397190475396e-05,
+      "loss": 1.4629,
+      "step": 104
+    },
+    {
+      "epoch": 0.016506514177916643,
+      "grad_norm": 0.07542011886835098,
+      "learning_rate": 4.999325443757309e-05,
+      "loss": 1.5976,
+      "step": 105
+    },
+    {
+      "epoch": 0.016663719074849183,
+      "grad_norm": 0.07440067082643509,
+      "learning_rate": 4.9993110158177895e-05,
+      "loss": 1.5469,
+      "step": 106
+    },
+    {
+      "epoch": 0.016820923971781723,
+      "grad_norm": 0.07547372579574585,
+      "learning_rate": 4.999296435229863e-05,
+      "loss": 1.5328,
+      "step": 107
+    },
+    {
+      "epoch": 0.01697812886871426,
+      "grad_norm": 0.07532137632369995,
+      "learning_rate": 4.999281701994419e-05,
+      "loss": 1.6742,
+      "step": 108
+    },
+    {
+      "epoch": 0.0171353337656468,
+      "grad_norm": 0.07249438762664795,
+      "learning_rate": 4.999266816112358e-05,
+      "loss": 1.4799,
+      "step": 109
+    },
+    {
+      "epoch": 0.01729253866257934,
+      "grad_norm": 0.07399806380271912,
+      "learning_rate": 4.999251777584589e-05,
+      "loss": 1.5438,
+      "step": 110
+    },
+    {
+      "epoch": 0.017449743559511878,
+      "grad_norm": 0.08135057240724564,
+      "learning_rate": 4.99923658641203e-05,
+      "loss": 1.5608,
+      "step": 111
+    },
+    {
+      "epoch": 0.017606948456444418,
+      "grad_norm": 0.07508935779333115,
+      "learning_rate": 4.99922124259561e-05,
+      "loss": 1.5894,
+      "step": 112
+    },
+    {
+      "epoch": 0.017764153353376957,
+      "grad_norm": 0.07432372123003006,
+      "learning_rate": 4.999205746136265e-05,
+      "loss": 1.4818,
+      "step": 113
+    },
+    {
+      "epoch": 0.017921358250309497,
+      "grad_norm": 0.07694194465875626,
+      "learning_rate": 4.999190097034942e-05,
+      "loss": 1.5629,
+      "step": 114
+    },
+    {
+      "epoch": 0.018078563147242037,
+      "grad_norm": 0.07384433597326279,
+      "learning_rate": 4.999174295292597e-05,
+      "loss": 1.4829,
+      "step": 115
+    },
+    {
+      "epoch": 0.018235768044174577,
+      "grad_norm": 0.07152919471263885,
+      "learning_rate": 4.999158340910195e-05,
+      "loss": 1.4748,
+      "step": 116
+    },
+    {
+      "epoch": 0.018392972941107116,
+      "grad_norm": 0.07719701528549194,
+      "learning_rate": 4.999142233888709e-05,
+      "loss": 1.5524,
+      "step": 117
+    },
+    {
+      "epoch": 0.018550177838039656,
+      "grad_norm": 0.07540587335824966,
+      "learning_rate": 4.999125974229125e-05,
+      "loss": 1.4661,
+      "step": 118
+    },
+    {
+      "epoch": 0.018707382734972196,
+      "grad_norm": 0.0787581130862236,
+      "learning_rate": 4.9991095619324344e-05,
+      "loss": 1.6455,
+      "step": 119
+    },
+    {
+      "epoch": 0.018864587631904736,
+      "grad_norm": 0.07454577833414078,
+      "learning_rate": 4.999092996999641e-05,
+      "loss": 1.5083,
+      "step": 120
+    },
+    {
+      "epoch": 0.019021792528837272,
+      "grad_norm": 0.0751076266169548,
+      "learning_rate": 4.9990762794317545e-05,
+      "loss": 1.4874,
+      "step": 121
+    },
+    {
+      "epoch": 0.01917899742576981,
+      "grad_norm": 0.07733119279146194,
+      "learning_rate": 4.999059409229798e-05,
+      "loss": 1.6308,
+      "step": 122
+    },
+    {
+      "epoch": 0.01933620232270235,
+      "grad_norm": 0.07897089421749115,
+      "learning_rate": 4.999042386394802e-05,
+      "loss": 1.5906,
+      "step": 123
+    },
+    {
+      "epoch": 0.01949340721963489,
+      "grad_norm": 0.07758141309022903,
+      "learning_rate": 4.999025210927804e-05,
+      "loss": 1.5604,
+      "step": 124
+    },
+    {
+      "epoch": 0.01965061211656743,
+      "grad_norm": 0.07845707982778549,
+      "learning_rate": 4.9990078828298544e-05,
+      "loss": 1.5901,
+      "step": 125
+    },
+    {
+      "epoch": 0.01980781701349997,
+      "grad_norm": 0.0772818773984909,
+      "learning_rate": 4.998990402102012e-05,
+      "loss": 1.4516,
+      "step": 126
+    },
+    {
+      "epoch": 0.01996502191043251,
+      "grad_norm": 0.07795504480600357,
+      "learning_rate": 4.998972768745344e-05,
+      "loss": 1.4642,
+      "step": 127
+    },
+    {
+      "epoch": 0.02012222680736505,
+      "grad_norm": 0.0784008800983429,
+      "learning_rate": 4.998954982760926e-05,
+      "loss": 1.5936,
+      "step": 128
+    },
+    {
+      "epoch": 0.02027943170429759,
+      "grad_norm": 0.07791212201118469,
+      "learning_rate": 4.9989370441498465e-05,
+      "loss": 1.4705,
+      "step": 129
+    },
+    {
+      "epoch": 0.02043663660123013,
+      "grad_norm": 0.07785367220640182,
+      "learning_rate": 4.9989189529132004e-05,
+      "loss": 1.5085,
+      "step": 130
+    },
+    {
+      "epoch": 0.02059384149816267,
+      "grad_norm": 0.07916689664125443,
+      "learning_rate": 4.9989007090520925e-05,
+      "loss": 1.5365,
+      "step": 131
+    },
+    {
+      "epoch": 0.02075104639509521,
+      "grad_norm": 0.0775083601474762,
+      "learning_rate": 4.9988823125676367e-05,
+      "loss": 1.5286,
+      "step": 132
+    },
+    {
+      "epoch": 0.020908251292027745,
+      "grad_norm": 0.08110442757606506,
+      "learning_rate": 4.998863763460956e-05,
+      "loss": 1.5779,
+      "step": 133
+    },
+    {
+      "epoch": 0.021065456188960285,
+      "grad_norm": 0.0814640000462532,
+      "learning_rate": 4.998845061733185e-05,
+      "loss": 1.4778,
+      "step": 134
+    },
+    {
+      "epoch": 0.021222661085892824,
+      "grad_norm": 0.08069492131471634,
+      "learning_rate": 4.998826207385465e-05,
+      "loss": 1.5317,
+      "step": 135
+    },
+    {
+      "epoch": 0.021379865982825364,
+      "grad_norm": 0.07377774268388748,
+      "learning_rate": 4.998807200418948e-05,
+      "loss": 1.5258,
+      "step": 136
+    },
+    {
+      "epoch": 0.021537070879757904,
+      "grad_norm": 0.0787922590970993,
+      "learning_rate": 4.9987880408347945e-05,
+      "loss": 1.5185,
+      "step": 137
+    },
+    {
+      "epoch": 0.021694275776690444,
+      "grad_norm": 0.07662995159626007,
+      "learning_rate": 4.9987687286341745e-05,
+      "loss": 1.4637,
+      "step": 138
+    },
+    {
+      "epoch": 0.021851480673622983,
+      "grad_norm": 0.08528955280780792,
+      "learning_rate": 4.9987492638182676e-05,
+      "loss": 1.4776,
+      "step": 139
+    },
+    {
+      "epoch": 0.022008685570555523,
+      "grad_norm": 0.08089053630828857,
+      "learning_rate": 4.9987296463882626e-05,
+      "loss": 1.5885,
+      "step": 140
+    },
+    {
+      "epoch": 0.022165890467488063,
+      "grad_norm": 0.08029694855213165,
+      "learning_rate": 4.998709876345358e-05,
+      "loss": 1.4557,
+      "step": 141
+    },
+    {
+      "epoch": 0.022323095364420602,
+      "grad_norm": 0.07918502390384674,
+      "learning_rate": 4.9986899536907614e-05,
+      "loss": 1.4285,
+      "step": 142
+    },
+    {
+      "epoch": 0.022480300261353142,
+      "grad_norm": 0.0813126415014267,
+      "learning_rate": 4.998669878425689e-05,
+      "loss": 1.5958,
+      "step": 143
+    },
+    {
+      "epoch": 0.022637505158285682,
+      "grad_norm": 0.07935188710689545,
+      "learning_rate": 4.998649650551368e-05,
+      "loss": 1.5249,
+      "step": 144
+    },
+    {
+      "epoch": 0.02279471005521822,
+      "grad_norm": 0.08163304626941681,
+      "learning_rate": 4.9986292700690324e-05,
+      "loss": 1.483,
+      "step": 145
+    },
+    {
+      "epoch": 0.022951914952150758,
+      "grad_norm": 0.08277447521686554,
+      "learning_rate": 4.998608736979928e-05,
+      "loss": 1.6212,
+      "step": 146
+    },
+    {
+      "epoch": 0.023109119849083298,
+      "grad_norm": 0.08285827934741974,
+      "learning_rate": 4.9985880512853076e-05,
+      "loss": 1.4495,
+      "step": 147
+    },
+    {
+      "epoch": 0.023266324746015837,
+      "grad_norm": 0.082750603556633,
+      "learning_rate": 4.998567212986437e-05,
+      "loss": 1.4335,
+      "step": 148
+    },
+    {
+      "epoch": 0.023423529642948377,
+      "grad_norm": 0.07986058294773102,
+      "learning_rate": 4.998546222084587e-05,
+      "loss": 1.4704,
+      "step": 149
+    },
+    {
+      "epoch": 0.023580734539880917,
+      "grad_norm": 0.08105576783418655,
+      "learning_rate": 4.9985250785810396e-05,
+      "loss": 1.5183,
+      "step": 150
+    },
+    {
+      "epoch": 0.023737939436813457,
+      "grad_norm": 0.08202917128801346,
+      "learning_rate": 4.9985037824770866e-05,
+      "loss": 1.5423,
+      "step": 151
+    },
+    {
+      "epoch": 0.023895144333745996,
+      "grad_norm": 0.08937894552946091,
+      "learning_rate": 4.998482333774029e-05,
+      "loss": 1.5731,
+      "step": 152
+    },
+    {
+      "epoch": 0.024052349230678536,
+      "grad_norm": 0.08333728462457657,
+      "learning_rate": 4.9984607324731766e-05,
+      "loss": 1.5133,
+      "step": 153
+    },
+    {
+      "epoch": 0.024209554127611076,
+      "grad_norm": 0.08529175072908401,
+      "learning_rate": 4.998438978575849e-05,
+      "loss": 1.516,
+      "step": 154
+    },
+    {
+      "epoch": 0.024366759024543615,
+      "grad_norm": 0.08508963882923126,
+      "learning_rate": 4.998417072083374e-05,
+      "loss": 1.5646,
+      "step": 155
+    },
+    {
+      "epoch": 0.024523963921476155,
+      "grad_norm": 0.08971578627824783,
+      "learning_rate": 4.99839501299709e-05,
+      "loss": 1.4714,
+      "step": 156
+    },
+    {
+      "epoch": 0.024681168818408695,
+      "grad_norm": 0.08380109816789627,
+      "learning_rate": 4.998372801318345e-05,
+      "loss": 1.4476,
+      "step": 157
+    },
+    {
+      "epoch": 0.024838373715341235,
+      "grad_norm": 0.08533143252134323,
+      "learning_rate": 4.9983504370484945e-05,
+      "loss": 1.4866,
+      "step": 158
+    },
+    {
+      "epoch": 0.02499557861227377,
+      "grad_norm": 0.08318709582090378,
+      "learning_rate": 4.998327920188905e-05,
+      "loss": 1.5274,
+      "step": 159
+    },
+    {
+      "epoch": 0.02515278350920631,
+      "grad_norm": 0.08486370742321014,
+      "learning_rate": 4.9983052507409525e-05,
+      "loss": 1.4713,
+      "step": 160
+    },
+    {
+      "epoch": 0.02515278350920631,
+      "eval_loss": 1.5136528015136719,
+      "eval_runtime": 2318.8971,
+      "eval_samples_per_second": 3.992,
+      "eval_steps_per_second": 1.996,
+      "step": 160
+    },
+    {
+      "epoch": 0.02530998840613885,
+      "grad_norm": 0.08242359757423401,
+      "learning_rate": 4.9982824287060195e-05,
+      "loss": 1.5069,
+      "step": 161
+    },
+    {
+      "epoch": 0.02546719330307139,
+      "grad_norm": 0.08547423779964447,
+      "learning_rate": 4.9982594540855014e-05,
+      "loss": 1.4973,
+      "step": 162
+    },
+    {
+      "epoch": 0.02562439820000393,
+      "grad_norm": 0.08345580101013184,
+      "learning_rate": 4.9982363268808016e-05,
+      "loss": 1.5078,
+      "step": 163
+    },
+    {
+      "epoch": 0.02578160309693647,
+      "grad_norm": 0.0830339640378952,
+      "learning_rate": 4.9982130470933316e-05,
+      "loss": 1.4098,
+      "step": 164
+    },
+    {
+      "epoch": 0.02593880799386901,
+      "grad_norm": 0.08568515628576279,
+      "learning_rate": 4.998189614724514e-05,
+      "loss": 1.4628,
+      "step": 165
+    },
+    {
+      "epoch": 0.02609601289080155,
+      "grad_norm": 0.08261829614639282,
+      "learning_rate": 4.998166029775779e-05,
+      "loss": 1.4492,
+      "step": 166
+    },
+    {
+      "epoch": 0.02625321778773409,
+      "grad_norm": 0.08944887667894363,
+      "learning_rate": 4.998142292248569e-05,
+      "loss": 1.5633,
+      "step": 167
+    },
+    {
+      "epoch": 0.02641042268466663,
+      "grad_norm": 0.08632911741733551,
+      "learning_rate": 4.998118402144332e-05,
+      "loss": 1.5106,
+      "step": 168
+    },
+    {
+      "epoch": 0.026567627581599168,
+      "grad_norm": 0.08733859658241272,
+      "learning_rate": 4.998094359464528e-05,
+      "loss": 1.5607,
+      "step": 169
+    },
+    {
+      "epoch": 0.026724832478531708,
+      "grad_norm": 0.08667927235364914,
+      "learning_rate": 4.9980701642106245e-05,
+      "loss": 1.4544,
+      "step": 170
+    },
+    {
+      "epoch": 0.026882037375464244,
+      "grad_norm": 0.08655022084712982,
+      "learning_rate": 4.9980458163841006e-05,
+      "loss": 1.5264,
+      "step": 171
+    },
+    {
+      "epoch": 0.027039242272396784,
+      "grad_norm": 0.08899988234043121,
+      "learning_rate": 4.9980213159864426e-05,
+      "loss": 1.4778,
+      "step": 172
+    },
+    {
+      "epoch": 0.027196447169329323,
+      "grad_norm": 0.09411856532096863,
+      "learning_rate": 4.997996663019147e-05,
+      "loss": 1.5269,
+      "step": 173
+    },
+    {
+      "epoch": 0.027353652066261863,
+      "grad_norm": 0.087191641330719,
+      "learning_rate": 4.997971857483719e-05,
+      "loss": 1.5166,
+      "step": 174
+    },
+    {
+      "epoch": 0.027510856963194403,
+      "grad_norm": 0.08959636092185974,
+      "learning_rate": 4.997946899381675e-05,
+      "loss": 1.5503,
+      "step": 175
+    },
+    {
+      "epoch": 0.027668061860126943,
+      "grad_norm": 0.0951187014579773,
+      "learning_rate": 4.997921788714537e-05,
+      "loss": 1.4879,
+      "step": 176
+    },
+    {
+      "epoch": 0.027825266757059482,
+      "grad_norm": 0.09324768930673599,
+      "learning_rate": 4.997896525483841e-05,
+      "loss": 1.5714,
+      "step": 177
+    },
+    {
+      "epoch": 0.027982471653992022,
+      "grad_norm": 0.08633986115455627,
+      "learning_rate": 4.997871109691129e-05,
+      "loss": 1.4198,
+      "step": 178
+    },
+    {
+      "epoch": 0.028139676550924562,
+      "grad_norm": 0.08947525173425674,
+      "learning_rate": 4.9978455413379535e-05,
+      "loss": 1.4702,
+      "step": 179
+    },
+    {
+      "epoch": 0.0282968814478571,
+      "grad_norm": 0.09275490790605545,
+      "learning_rate": 4.9978198204258766e-05,
+      "loss": 1.5252,
+      "step": 180
+    },
+    {
+      "epoch": 0.02845408634478964,
+      "grad_norm": 0.08761609345674515,
+      "learning_rate": 4.9977939469564676e-05,
+      "loss": 1.505,
+      "step": 181
+    },
+    {
+      "epoch": 0.02861129124172218,
+      "grad_norm": 0.08683087676763535,
+      "learning_rate": 4.997767920931308e-05,
+      "loss": 1.5059,
+      "step": 182
+    },
+    {
+      "epoch": 0.02876849613865472,
+      "grad_norm": 0.08931361883878708,
+      "learning_rate": 4.997741742351988e-05,
+      "loss": 1.5003,
+      "step": 183
+    },
+    {
+      "epoch": 0.028925701035587257,
+      "grad_norm": 0.08820109069347382,
+      "learning_rate": 4.997715411220105e-05,
+      "loss": 1.5132,
+      "step": 184
+    },
+    {
+      "epoch": 0.029082905932519797,
+      "grad_norm": 0.09284964948892593,
+      "learning_rate": 4.997688927537268e-05,
+      "loss": 1.4561,
+      "step": 185
+    },
+    {
+      "epoch": 0.029240110829452336,
+      "grad_norm": 0.09472864121198654,
+      "learning_rate": 4.997662291305094e-05,
+      "loss": 1.4729,
+      "step": 186
+    },
+    {
+      "epoch": 0.029397315726384876,
+      "grad_norm": 0.08725330233573914,
+      "learning_rate": 4.997635502525211e-05,
+      "loss": 1.3994,
+      "step": 187
+    },
+    {
+      "epoch": 0.029554520623317416,
+      "grad_norm": 0.09085626900196075,
+      "learning_rate": 4.9976085611992536e-05,
+      "loss": 1.4695,
+      "step": 188
+    },
+    {
+      "epoch": 0.029711725520249956,
+      "grad_norm": 0.09322400391101837,
+      "learning_rate": 4.9975814673288684e-05,
+      "loss": 1.4753,
+      "step": 189
+    },
+    {
+      "epoch": 0.029868930417182495,
+      "grad_norm": 0.08927160501480103,
+      "learning_rate": 4.99755422091571e-05,
+      "loss": 1.4465,
+      "step": 190
+    },
+    {
+      "epoch": 0.030026135314115035,
+      "grad_norm": 0.09317070990800858,
+      "learning_rate": 4.997526821961442e-05,
+      "loss": 1.5124,
+      "step": 191
+    },
+    {
+      "epoch": 0.030183340211047575,
+      "grad_norm": 0.08911167085170746,
+      "learning_rate": 4.9974992704677385e-05,
+      "loss": 1.4515,
+      "step": 192
+    },
+    {
+      "epoch": 0.030340545107980114,
+      "grad_norm": 0.09432853013277054,
+      "learning_rate": 4.997471566436282e-05,
+      "loss": 1.4623,
+      "step": 193
+    },
+    {
+      "epoch": 0.030497750004912654,
+      "grad_norm": 0.09417332708835602,
+      "learning_rate": 4.997443709868764e-05,
+      "loss": 1.5103,
+      "step": 194
+    },
+    {
+      "epoch": 0.030654954901845194,
+      "grad_norm": 0.09564542025327682,
+      "learning_rate": 4.997415700766887e-05,
+      "loss": 1.4929,
+      "step": 195
+    },
+    {
+      "epoch": 0.030812159798777734,
+      "grad_norm": 0.09101004898548126,
+      "learning_rate": 4.997387539132361e-05,
+      "loss": 1.4225,
+      "step": 196
+    },
+    {
+      "epoch": 0.03096936469571027,
+      "grad_norm": 0.09196274727582932,
+      "learning_rate": 4.997359224966906e-05,
+      "loss": 1.4701,
+      "step": 197
+    },
+    {
+      "epoch": 0.03112656959264281,
+      "grad_norm": 0.09573279321193695,
+      "learning_rate": 4.997330758272251e-05,
+      "loss": 1.4425,
+      "step": 198
+    },
+    {
+      "epoch": 0.03128377448957535,
+      "grad_norm": 0.09180758893489838,
+      "learning_rate": 4.9973021390501354e-05,
+      "loss": 1.4426,
+      "step": 199
+    },
+    {
+      "epoch": 0.03144097938650789,
+      "grad_norm": 0.09583238512277603,
+      "learning_rate": 4.997273367302306e-05,
+      "loss": 1.5158,
+      "step": 200
+    },
+    {
+      "epoch": 0.03159818428344043,
+      "grad_norm": 0.09394747018814087,
+      "learning_rate": 4.997244443030521e-05,
+      "loss": 1.4306,
+      "step": 201
+    },
+    {
+      "epoch": 0.03175538918037297,
+      "grad_norm": 0.09470199793577194,
+      "learning_rate": 4.9972153662365474e-05,
+      "loss": 1.5286,
+      "step": 202
+    },
+    {
+      "epoch": 0.031912594077305505,
+      "grad_norm": 0.09274959564208984,
+      "learning_rate": 4.997186136922161e-05,
+      "loss": 1.4803,
+      "step": 203
+    },
+    {
+      "epoch": 0.03206979897423805,
+      "grad_norm": 0.09344369918107986,
+      "learning_rate": 4.997156755089145e-05,
+      "loss": 1.5449,
+      "step": 204
+    },
+    {
+      "epoch": 0.032227003871170584,
+      "grad_norm": 0.09794919937849045,
+      "learning_rate": 4.997127220739296e-05,
+      "loss": 1.4383,
+      "step": 205
+    },
+    {
+      "epoch": 0.03238420876810313,
+      "grad_norm": 0.09698093682527542,
+      "learning_rate": 4.997097533874418e-05,
+      "loss": 1.4462,
+      "step": 206
+    },
+    {
+      "epoch": 0.032541413665035664,
+      "grad_norm": 0.09690559655427933,
+      "learning_rate": 4.997067694496323e-05,
+      "loss": 1.4735,
+      "step": 207
+    },
+    {
+      "epoch": 0.03269861856196821,
+      "grad_norm": 0.09657544642686844,
+      "learning_rate": 4.9970377026068336e-05,
+      "loss": 1.5672,
+      "step": 208
+    },
+    {
+      "epoch": 0.03285582345890074,
+      "grad_norm": 0.09483659267425537,
+      "learning_rate": 4.9970075582077825e-05,
+      "loss": 1.4931,
+      "step": 209
+    },
+    {
+      "epoch": 0.033013028355833286,
+      "grad_norm": 0.09744243323802948,
+      "learning_rate": 4.9969772613010104e-05,
+      "loss": 1.4638,
+      "step": 210
+    },
+    {
+      "epoch": 0.03317023325276582,
+      "grad_norm": 0.09521006047725677,
+      "learning_rate": 4.9969468118883665e-05,
+      "loss": 1.4127,
+      "step": 211
+    },
+    {
+      "epoch": 0.033327438149698366,
+      "grad_norm": 0.09646004438400269,
+      "learning_rate": 4.996916209971713e-05,
+      "loss": 1.5139,
+      "step": 212
+    },
+    {
+      "epoch": 0.0334846430466309,
+      "grad_norm": 0.09292810410261154,
+      "learning_rate": 4.996885455552916e-05,
+      "loss": 1.4399,
+      "step": 213
+    },
+    {
+      "epoch": 0.033641847943563445,
+      "grad_norm": 0.09986516088247299,
+      "learning_rate": 4.996854548633857e-05,
+      "loss": 1.4637,
+      "step": 214
+    },
+    {
+      "epoch": 0.03379905284049598,
+      "grad_norm": 0.09723702073097229,
+      "learning_rate": 4.996823489216421e-05,
+      "loss": 1.5673,
+      "step": 215
+    },
+    {
+      "epoch": 0.03395625773742852,
+      "grad_norm": 0.09608977288007736,
+      "learning_rate": 4.996792277302507e-05,
+      "loss": 1.4428,
+      "step": 216
+    },
+    {
+      "epoch": 0.03411346263436106,
+      "grad_norm": 0.09329380095005035,
+      "learning_rate": 4.99676091289402e-05,
+      "loss": 1.3892,
+      "step": 217
+    },
+    {
+      "epoch": 0.0342706675312936,
+      "grad_norm": 0.0959913358092308,
+      "learning_rate": 4.996729395992875e-05,
+      "loss": 1.5219,
+      "step": 218
+    },
+    {
+      "epoch": 0.03442787242822614,
+      "grad_norm": 0.09832671284675598,
+      "learning_rate": 4.996697726600999e-05,
+      "loss": 1.5259,
+      "step": 219
+    },
+    {
+      "epoch": 0.03458507732515868,
+      "grad_norm": 0.10061636567115784,
+      "learning_rate": 4.996665904720325e-05,
+      "loss": 1.5216,
+      "step": 220
+    },
+    {
+      "epoch": 0.03474228222209122,
+      "grad_norm": 0.09742400050163269,
+      "learning_rate": 4.9966339303527965e-05,
+      "loss": 1.3819,
+      "step": 221
+    },
+    {
+      "epoch": 0.034899487119023756,
+      "grad_norm": 0.09629969298839569,
+      "learning_rate": 4.996601803500367e-05,
+      "loss": 1.5341,
+      "step": 222
+    },
+    {
+      "epoch": 0.0350566920159563,
+      "grad_norm": 0.09776200354099274,
+      "learning_rate": 4.996569524164998e-05,
+      "loss": 1.5054,
+      "step": 223
+    },
+    {
+      "epoch": 0.035213896912888835,
+      "grad_norm": 0.1008530780673027,
+      "learning_rate": 4.996537092348661e-05,
+      "loss": 1.5333,
+      "step": 224
+    },
+    {
+      "epoch": 0.03537110180982138,
+      "grad_norm": 0.09749735891819,
+      "learning_rate": 4.996504508053338e-05,
+      "loss": 1.3899,
+      "step": 225
+    },
+    {
+      "epoch": 0.035528306706753915,
+      "grad_norm": 0.10522401332855225,
+      "learning_rate": 4.9964717712810175e-05,
+      "loss": 1.5413,
+      "step": 226
+    },
+    {
+      "epoch": 0.03568551160368646,
+      "grad_norm": 0.09566272795200348,
+      "learning_rate": 4.9964388820336996e-05,
+      "loss": 1.435,
+      "step": 227
+    },
+    {
+      "epoch": 0.035842716500618994,
+      "grad_norm": 0.10133984684944153,
+      "learning_rate": 4.996405840313393e-05,
+      "loss": 1.445,
+      "step": 228
+    },
+    {
+      "epoch": 0.03599992139755153,
+      "grad_norm": 0.09702739119529724,
+      "learning_rate": 4.996372646122116e-05,
+      "loss": 1.4287,
+      "step": 229
+    },
+    {
+      "epoch": 0.036157126294484074,
+      "grad_norm": 0.1012992411851883,
+      "learning_rate": 4.996339299461896e-05,
+      "loss": 1.382,
+      "step": 230
+    },
+    {
+      "epoch": 0.03631433119141661,
+      "grad_norm": 0.09877166152000427,
+      "learning_rate": 4.99630580033477e-05,
+      "loss": 1.5729,
+      "step": 231
+    },
+    {
+      "epoch": 0.03647153608834915,
+      "grad_norm": 0.1033129170536995,
+      "learning_rate": 4.996272148742783e-05,
+      "loss": 1.4754,
+      "step": 232
+    },
+    {
+      "epoch": 0.03662874098528169,
+      "grad_norm": 0.09901215881109238,
+      "learning_rate": 4.9962383446879914e-05,
+      "loss": 1.5153,
+      "step": 233
+    },
+    {
+      "epoch": 0.03678594588221423,
+      "grad_norm": 0.10241983830928802,
+      "learning_rate": 4.996204388172458e-05,
+      "loss": 1.5131,
+      "step": 234
+    },
+    {
+      "epoch": 0.03694315077914677,
+      "grad_norm": 0.09574593603610992,
+      "learning_rate": 4.9961702791982594e-05,
+      "loss": 1.5285,
+      "step": 235
+    },
+    {
+      "epoch": 0.03710035567607931,
+      "grad_norm": 0.10309838503599167,
+      "learning_rate": 4.996136017767477e-05,
+      "loss": 1.5751,
+      "step": 236
+    },
+    {
+      "epoch": 0.03725756057301185,
+      "grad_norm": 0.09928470849990845,
+      "learning_rate": 4.996101603882204e-05,
+      "loss": 1.5108,
+      "step": 237
+    },
+    {
+      "epoch": 0.03741476546994439,
+      "grad_norm": 0.10514767467975616,
+      "learning_rate": 4.996067037544542e-05,
+      "loss": 1.4206,
+      "step": 238
+    },
+    {
+      "epoch": 0.03757197036687693,
+      "grad_norm": 0.10411518812179565,
+      "learning_rate": 4.996032318756601e-05,
+      "loss": 1.5628,
+      "step": 239
+    },
+    {
+      "epoch": 0.03772917526380947,
+      "grad_norm": 0.0989808738231659,
+      "learning_rate": 4.9959974475205045e-05,
+      "loss": 1.4444,
+      "step": 240
+    },
+    {
+      "epoch": 0.03788638016074201,
+      "grad_norm": 0.10069911926984787,
+      "learning_rate": 4.9959624238383804e-05,
+      "loss": 1.4805,
+      "step": 241
+    },
+    {
+      "epoch": 0.038043585057674544,
+      "grad_norm": 0.10637518763542175,
+      "learning_rate": 4.995927247712367e-05,
+      "loss": 1.5289,
+      "step": 242
+    },
+    {
+      "epoch": 0.03820078995460709,
+      "grad_norm": 0.10085684061050415,
+      "learning_rate": 4.995891919144614e-05,
+      "loss": 1.5288,
+      "step": 243
+    },
+    {
+      "epoch": 0.03835799485153962,
+      "grad_norm": 0.09989017248153687,
+      "learning_rate": 4.995856438137279e-05,
+      "loss": 1.5444,
+      "step": 244
+    },
+    {
+      "epoch": 0.038515199748472166,
+      "grad_norm": 0.10382463037967682,
+      "learning_rate": 4.9958208046925294e-05,
+      "loss": 1.4621,
+      "step": 245
+    },
+    {
+      "epoch": 0.0386724046454047,
+      "grad_norm": 0.10208063572645187,
+      "learning_rate": 4.99578501881254e-05,
+      "loss": 1.5003,
+      "step": 246
+    },
+    {
+      "epoch": 0.038829609542337246,
+      "grad_norm": 0.1028011366724968,
+      "learning_rate": 4.9957490804994977e-05,
+      "loss": 1.516,
+      "step": 247
+    },
+    {
+      "epoch": 0.03898681443926978,
+      "grad_norm": 0.10475701838731766,
+      "learning_rate": 4.995712989755598e-05,
+      "loss": 1.5333,
+      "step": 248
+    },
+    {
+      "epoch": 0.039144019336202325,
+      "grad_norm": 0.1038154736161232,
+      "learning_rate": 4.995676746583044e-05,
+      "loss": 1.4779,
+      "step": 249
+    },
+    {
+      "epoch": 0.03930122423313486,
+      "grad_norm": 0.10413440316915512,
+      "learning_rate": 4.99564035098405e-05,
+      "loss": 1.5241,
+      "step": 250
+    },
+    {
+      "epoch": 0.039458429130067404,
+      "grad_norm": 0.09869382530450821,
+      "learning_rate": 4.995603802960838e-05,
+      "loss": 1.442,
+      "step": 251
+    },
+    {
+      "epoch": 0.03961563402699994,
+      "grad_norm": 0.10138234496116638,
+      "learning_rate": 4.995567102515641e-05,
+      "loss": 1.5393,
+      "step": 252
+    },
+    {
+      "epoch": 0.039772838923932484,
+      "grad_norm": 0.10225867480039597,
+      "learning_rate": 4.995530249650701e-05,
+      "loss": 1.4516,
+      "step": 253
+    },
+    {
+      "epoch": 0.03993004382086502,
+      "grad_norm": 0.09942895174026489,
+      "learning_rate": 4.995493244368268e-05,
+      "loss": 1.4543,
+      "step": 254
+    },
+    {
+      "epoch": 0.040087248717797556,
+      "grad_norm": 0.11218860745429993,
+      "learning_rate": 4.995456086670602e-05,
+      "loss": 1.4985,
+      "step": 255
+    },
+    {
+      "epoch": 0.0402444536147301,
+      "grad_norm": 0.10839337855577469,
+      "learning_rate": 4.9954187765599736e-05,
+      "loss": 1.4805,
+      "step": 256
+    },
+    {
+      "epoch": 0.040401658511662636,
+      "grad_norm": 0.10317599028348923,
+      "learning_rate": 4.9953813140386595e-05,
+      "loss": 1.4412,
+      "step": 257
+    },
+    {
+      "epoch": 0.04055886340859518,
+      "grad_norm": 0.10285656154155731,
+      "learning_rate": 4.99534369910895e-05,
+      "loss": 1.476,
+      "step": 258
+    },
+    {
+      "epoch": 0.040716068305527715,
+      "grad_norm": 0.10330680012702942,
+      "learning_rate": 4.995305931773141e-05,
+      "loss": 1.5157,
+      "step": 259
+    },
+    {
+      "epoch": 0.04087327320246026,
+      "grad_norm": 0.1086694598197937,
+      "learning_rate": 4.99526801203354e-05,
+      "loss": 1.4999,
+      "step": 260
+    },
+    {
+      "epoch": 0.041030478099392795,
+      "grad_norm": 0.10800144821405411,
+      "learning_rate": 4.995229939892464e-05,
+      "loss": 1.4764,
+      "step": 261
+    },
+    {
+      "epoch": 0.04118768299632534,
+      "grad_norm": 0.10645303875207901,
+      "learning_rate": 4.9951917153522355e-05,
+      "loss": 1.4404,
+      "step": 262
+    },
+    {
+      "epoch": 0.041344887893257874,
+      "grad_norm": 0.10440964996814728,
+      "learning_rate": 4.9951533384151906e-05,
+      "loss": 1.3678,
+      "step": 263
+    },
+    {
+      "epoch": 0.04150209279019042,
+      "grad_norm": 0.10993078351020813,
+      "learning_rate": 4.995114809083673e-05,
+      "loss": 1.5064,
+      "step": 264
+    },
+    {
+      "epoch": 0.041659297687122954,
+      "grad_norm": 0.10710245370864868,
+      "learning_rate": 4.9950761273600366e-05,
+      "loss": 1.4134,
+      "step": 265
+    },
+    {
+      "epoch": 0.04181650258405549,
+      "grad_norm": 0.11030582338571548,
+      "learning_rate": 4.995037293246644e-05,
+      "loss": 1.5299,
+      "step": 266
+    },
+    {
+      "epoch": 0.04197370748098803,
+      "grad_norm": 0.1058267131447792,
+      "learning_rate": 4.994998306745866e-05,
+      "loss": 1.3654,
+      "step": 267
+    },
+    {
+      "epoch": 0.04213091237792057,
+      "grad_norm": 0.10541702806949615,
+      "learning_rate": 4.994959167860084e-05,
+      "loss": 1.4297,
+      "step": 268
+    },
+    {
+      "epoch": 0.04228811727485311,
+      "grad_norm": 0.11085420846939087,
+      "learning_rate": 4.994919876591689e-05,
+      "loss": 1.4876,
+      "step": 269
+    },
+    {
+      "epoch": 0.04244532217178565,
+      "grad_norm": 0.11054470390081406,
+      "learning_rate": 4.994880432943081e-05,
+      "loss": 1.574,
+      "step": 270
+    },
+    {
+      "epoch": 0.04260252706871819,
+      "grad_norm": 0.11234510689973831,
+      "learning_rate": 4.994840836916668e-05,
+      "loss": 1.5079,
+      "step": 271
+    },
+    {
+      "epoch": 0.04275973196565073,
+      "grad_norm": 0.11040106415748596,
+      "learning_rate": 4.994801088514869e-05,
+      "loss": 1.5091,
+      "step": 272
+    },
+    {
+      "epoch": 0.04291693686258327,
+      "grad_norm": 0.10639887303113937,
+      "learning_rate": 4.994761187740111e-05,
+      "loss": 1.4495,
+      "step": 273
+    },
+    {
+      "epoch": 0.04307414175951581,
+      "grad_norm": 0.11268071085214615,
+      "learning_rate": 4.994721134594833e-05,
+      "loss": 1.5057,
+      "step": 274
+    },
+    {
+      "epoch": 0.04323134665644835,
+      "grad_norm": 0.10079260170459747,
+      "learning_rate": 4.994680929081479e-05,
+      "loss": 1.4145,
+      "step": 275
+    },
+    {
+      "epoch": 0.04338855155338089,
+      "grad_norm": 0.11474710702896118,
+      "learning_rate": 4.994640571202506e-05,
+      "loss": 1.5061,
+      "step": 276
+    },
+    {
+      "epoch": 0.04354575645031343,
+      "grad_norm": 0.10946876555681229,
+      "learning_rate": 4.994600060960377e-05,
+      "loss": 1.5306,
+      "step": 277
+    },
+    {
+      "epoch": 0.04370296134724597,
+      "grad_norm": 0.11192137002944946,
+      "learning_rate": 4.994559398357569e-05,
+      "loss": 1.5347,
+      "step": 278
+    },
+    {
+      "epoch": 0.0438601662441785,
+      "grad_norm": 0.10744784027338028,
+      "learning_rate": 4.994518583396564e-05,
+      "loss": 1.4686,
+      "step": 279
+    },
+    {
+      "epoch": 0.044017371141111046,
+      "grad_norm": 0.11113352328538895,
+      "learning_rate": 4.9944776160798544e-05,
+      "loss": 1.4101,
+      "step": 280
+    },
+    {
+      "epoch": 0.04417457603804358,
+      "grad_norm": 0.11456230282783508,
+      "learning_rate": 4.994436496409943e-05,
+      "loss": 1.4036,
+      "step": 281
+    },
+    {
+      "epoch": 0.044331780934976125,
+      "grad_norm": 0.11608672887086868,
+      "learning_rate": 4.994395224389342e-05,
+      "loss": 1.4949,
+      "step": 282
+    },
+    {
+      "epoch": 0.04448898583190866,
+      "grad_norm": 0.1232326403260231,
+      "learning_rate": 4.9943538000205705e-05,
+      "loss": 1.5501,
+      "step": 283
+    },
+    {
+      "epoch": 0.044646190728841205,
+      "grad_norm": 0.11791515350341797,
+      "learning_rate": 4.994312223306159e-05,
+      "loss": 1.4542,
+      "step": 284
+    },
+    {
+      "epoch": 0.04480339562577374,
+      "grad_norm": 0.11657550930976868,
+      "learning_rate": 4.9942704942486476e-05,
+      "loss": 1.4724,
+      "step": 285
+    },
+    {
+      "epoch": 0.044960600522706284,
+      "grad_norm": 0.11560262739658356,
+      "learning_rate": 4.994228612850584e-05,
+      "loss": 1.4036,
+      "step": 286
+    },
+    {
+      "epoch": 0.04511780541963882,
+      "grad_norm": 0.10999175906181335,
+      "learning_rate": 4.994186579114527e-05,
+      "loss": 1.4489,
+      "step": 287
+    },
+    {
+      "epoch": 0.045275010316571364,
+      "grad_norm": 0.11586826294660568,
+      "learning_rate": 4.9941443930430436e-05,
+      "loss": 1.5486,
+      "step": 288
+    },
+    {
+      "epoch": 0.0454322152135039,
+      "grad_norm": 0.11349951475858688,
+      "learning_rate": 4.994102054638711e-05,
+      "loss": 1.5698,
+      "step": 289
+    },
+    {
+      "epoch": 0.04558942011043644,
+      "grad_norm": 0.11978698521852493,
+      "learning_rate": 4.9940595639041134e-05,
+      "loss": 1.3933,
+      "step": 290
+    },
+    {
+      "epoch": 0.04574662500736898,
+      "grad_norm": 0.11438622325658798,
+      "learning_rate": 4.994016920841846e-05,
+      "loss": 1.5005,
+      "step": 291
+    },
+    {
+      "epoch": 0.045903829904301516,
+      "grad_norm": 0.11395915597677231,
+      "learning_rate": 4.9939741254545155e-05,
+      "loss": 1.4521,
+      "step": 292
+    },
+    {
+      "epoch": 0.04606103480123406,
+      "grad_norm": 0.11659599095582962,
+      "learning_rate": 4.993931177744734e-05,
+      "loss": 1.5166,
+      "step": 293
+    },
+    {
+      "epoch": 0.046218239698166595,
+      "grad_norm": 0.11053171753883362,
+      "learning_rate": 4.9938880777151254e-05,
+      "loss": 1.4459,
+      "step": 294
+    },
+    {
+      "epoch": 0.04637544459509914,
+      "grad_norm": 0.11428084224462509,
+      "learning_rate": 4.993844825368321e-05,
+      "loss": 1.4448,
+      "step": 295
+    },
+    {
+      "epoch": 0.046532649492031675,
+      "grad_norm": 0.10734150558710098,
+      "learning_rate": 4.993801420706964e-05,
+      "loss": 1.3388,
+      "step": 296
+    },
+    {
+      "epoch": 0.04668985438896422,
+      "grad_norm": 0.11137369275093079,
+      "learning_rate": 4.993757863733703e-05,
+      "loss": 1.4155,
+      "step": 297
+    },
+    {
+      "epoch": 0.046847059285896754,
+      "grad_norm": 0.1221408098936081,
+      "learning_rate": 4.993714154451202e-05,
+      "loss": 1.4884,
+      "step": 298
+    },
+    {
+      "epoch": 0.0470042641828293,
+      "grad_norm": 0.11707969009876251,
+      "learning_rate": 4.993670292862127e-05,
+      "loss": 1.4605,
+      "step": 299
+    },
+    {
+      "epoch": 0.047161469079761834,
+      "grad_norm": 0.11751751601696014,
+      "learning_rate": 4.993626278969158e-05,
+      "loss": 1.5538,
+      "step": 300
+    },
+    {
+      "epoch": 0.04731867397669438,
+      "grad_norm": 0.11617731302976608,
+      "learning_rate": 4.993582112774984e-05,
+      "loss": 1.438,
+      "step": 301
+    },
+    {
+      "epoch": 0.04747587887362691,
+      "grad_norm": 0.15164637565612793,
+      "learning_rate": 4.993537794282302e-05,
+      "loss": 1.4607,
+      "step": 302
+    },
+    {
+      "epoch": 0.047633083770559456,
+      "grad_norm": 0.12434446811676025,
+      "learning_rate": 4.9934933234938193e-05,
+      "loss": 1.4167,
+      "step": 303
+    },
+    {
+      "epoch": 0.04779028866749199,
+      "grad_norm": 0.12518739700317383,
+      "learning_rate": 4.993448700412251e-05,
+      "loss": 1.4003,
+      "step": 304
+    },
+    {
+      "epoch": 0.04794749356442453,
+      "grad_norm": 0.11146944761276245,
+      "learning_rate": 4.993403925040323e-05,
+      "loss": 1.3913,
+      "step": 305
+    },
+    {
+      "epoch": 0.04810469846135707,
+      "grad_norm": 0.11682326346635818,
+      "learning_rate": 4.993358997380771e-05,
+      "loss": 1.3415,
+      "step": 306
+    },
+    {
+      "epoch": 0.04826190335828961,
+      "grad_norm": 0.1197504773736,
+      "learning_rate": 4.993313917436336e-05,
+      "loss": 1.515,
+      "step": 307
+    },
+    {
+      "epoch": 0.04841910825522215,
+      "grad_norm": 0.14647473394870758,
+      "learning_rate": 4.993268685209775e-05,
+      "loss": 1.4529,
+      "step": 308
+    },
+    {
+      "epoch": 0.04857631315215469,
+      "grad_norm": 0.12431525439023972,
+      "learning_rate": 4.9932233007038484e-05,
+      "loss": 1.5426,
+      "step": 309
+    },
+    {
+      "epoch": 0.04873351804908723,
+      "grad_norm": 0.11715538799762726,
+      "learning_rate": 4.9931777639213284e-05,
+      "loss": 1.4615,
+      "step": 310
+    },
+    {
+      "epoch": 0.04889072294601977,
+      "grad_norm": 0.12391876429319382,
+      "learning_rate": 4.993132074864997e-05,
+      "loss": 1.4138,
+      "step": 311
+    },
+    {
+      "epoch": 0.04904792784295231,
+      "grad_norm": 0.11894181370735168,
+      "learning_rate": 4.9930862335376444e-05,
+      "loss": 1.4383,
+      "step": 312
+    },
+    {
+      "epoch": 0.049205132739884846,
+      "grad_norm": 0.1225295439362526,
+      "learning_rate": 4.9930402399420695e-05,
+      "loss": 1.3847,
+      "step": 313
+    },
+    {
+      "epoch": 0.04936233763681739,
+      "grad_norm": 0.11435995995998383,
+      "learning_rate": 4.9929940940810825e-05,
+      "loss": 1.4254,
+      "step": 314
+    },
+    {
+      "epoch": 0.049519542533749926,
+      "grad_norm": 0.11988761276006699,
+      "learning_rate": 4.9929477959575024e-05,
+      "loss": 1.4787,
+      "step": 315
+    },
+    {
+      "epoch": 0.04967674743068247,
+      "grad_norm": 0.11983373016119003,
+      "learning_rate": 4.992901345574155e-05,
+      "loss": 1.4341,
+      "step": 316
+    },
+    {
+      "epoch": 0.049833952327615005,
+      "grad_norm": 0.13395054638385773,
+      "learning_rate": 4.992854742933878e-05,
+      "loss": 1.4315,
+      "step": 317
+    },
+    {
+      "epoch": 0.04999115722454754,
+      "grad_norm": 0.12578143179416656,
+      "learning_rate": 4.9928079880395186e-05,
+      "loss": 1.4143,
+      "step": 318
+    },
+    {
+      "epoch": 0.050148362121480085,
+      "grad_norm": 0.1401878446340561,
+      "learning_rate": 4.992761080893932e-05,
+      "loss": 1.4665,
+      "step": 319
+    },
+    {
+      "epoch": 0.05030556701841262,
+      "grad_norm": 0.13048145174980164,
+      "learning_rate": 4.9927140214999826e-05,
+      "loss": 1.4266,
+      "step": 320
+    },
+    {
+      "epoch": 0.05030556701841262,
+      "eval_loss": 1.450086236000061,
+      "eval_runtime": 2316.1877,
+      "eval_samples_per_second": 3.997,
+      "eval_steps_per_second": 1.999,
+      "step": 320
+    },
+    {
+      "epoch": 0.050462771915345164,
+      "grad_norm": 0.13121232390403748,
+      "learning_rate": 4.992666809860545e-05,
+      "loss": 1.4946,
+      "step": 321
+    },
+    {
+      "epoch": 0.0506199768122777,
+      "grad_norm": 0.13547195494174957,
+      "learning_rate": 4.9926194459785015e-05,
+      "loss": 1.5532,
+      "step": 322
+    },
+    {
+      "epoch": 0.050777181709210244,
+      "grad_norm": 0.11797169595956802,
+      "learning_rate": 4.992571929856747e-05,
+      "loss": 1.4118,
+      "step": 323
+    },
+    {
+      "epoch": 0.05093438660614278,
+      "grad_norm": 0.12734922766685486,
+      "learning_rate": 4.992524261498183e-05,
+      "loss": 1.4427,
+      "step": 324
+    },
+    {
+      "epoch": 0.05109159150307532,
+      "grad_norm": 0.12444902211427689,
+      "learning_rate": 4.99247644090572e-05,
+      "loss": 1.4369,
+      "step": 325
+    },
+    {
+      "epoch": 0.05124879640000786,
+      "grad_norm": 0.12244518846273422,
+      "learning_rate": 4.99242846808228e-05,
+      "loss": 1.4587,
+      "step": 326
+    },
+    {
+      "epoch": 0.0514060012969404,
+      "grad_norm": 0.12424397468566895,
+      "learning_rate": 4.9923803430307916e-05,
+      "loss": 1.3949,
+      "step": 327
+    },
+    {
+      "epoch": 0.05156320619387294,
+      "grad_norm": 0.1352718621492386,
+      "learning_rate": 4.9923320657541944e-05,
+      "loss": 1.504,
+      "step": 328
+    },
+    {
+      "epoch": 0.05172041109080548,
+      "grad_norm": 0.12855666875839233,
+      "learning_rate": 4.992283636255438e-05,
+      "loss": 1.4271,
+      "step": 329
+    },
+    {
+      "epoch": 0.05187761598773802,
+      "grad_norm": 0.129829540848732,
+      "learning_rate": 4.99223505453748e-05,
+      "loss": 1.455,
+      "step": 330
+    },
+    {
+      "epoch": 0.052034820884670555,
+      "grad_norm": 0.12780050933361053,
+      "learning_rate": 4.992186320603286e-05,
+      "loss": 1.4045,
+      "step": 331
+    },
+    {
+      "epoch": 0.0521920257816031,
+      "grad_norm": 0.13515712320804596,
+      "learning_rate": 4.992137434455834e-05,
+      "loss": 1.4335,
+      "step": 332
+    },
+    {
+      "epoch": 0.052349230678535634,
+      "grad_norm": 0.15026766061782837,
+      "learning_rate": 4.99208839609811e-05,
+      "loss": 1.5386,
+      "step": 333
+    },
+    {
+      "epoch": 0.05250643557546818,
+      "grad_norm": 0.13422101736068726,
+      "learning_rate": 4.992039205533108e-05,
+      "loss": 1.454,
+      "step": 334
+    },
+    {
+      "epoch": 0.05266364047240071,
+      "grad_norm": 0.13735777139663696,
+      "learning_rate": 4.991989862763833e-05,
+      "loss": 1.4415,
+      "step": 335
+    },
+    {
+      "epoch": 0.05282084536933326,
+      "grad_norm": 0.12985137104988098,
+      "learning_rate": 4.9919403677932994e-05,
+      "loss": 1.385,
+      "step": 336
+    },
+    {
+      "epoch": 0.05297805026626579,
+      "grad_norm": 0.1301167607307434,
+      "learning_rate": 4.9918907206245285e-05,
+      "loss": 1.4364,
+      "step": 337
+    },
+    {
+      "epoch": 0.053135255163198336,
+      "grad_norm": 0.1407599002122879,
+      "learning_rate": 4.991840921260553e-05,
+      "loss": 1.4454,
+      "step": 338
+    },
+    {
+      "epoch": 0.05329246006013087,
+      "grad_norm": 0.12763133645057678,
+      "learning_rate": 4.9917909697044164e-05,
+      "loss": 1.4008,
+      "step": 339
+    },
+    {
+      "epoch": 0.053449664957063416,
+      "grad_norm": 0.1443052589893341,
+      "learning_rate": 4.991740865959167e-05,
+      "loss": 1.5184,
+      "step": 340
+    },
+    {
+      "epoch": 0.05360686985399595,
+      "grad_norm": 0.13496418297290802,
+      "learning_rate": 4.991690610027866e-05,
+      "loss": 1.3888,
+      "step": 341
+    },
+    {
+      "epoch": 0.05376407475092849,
+      "grad_norm": 0.12681293487548828,
+      "learning_rate": 4.991640201913583e-05,
+      "loss": 1.42,
+      "step": 342
+    },
+    {
+      "epoch": 0.05392127964786103,
+      "grad_norm": 0.13178062438964844,
+      "learning_rate": 4.9915896416193965e-05,
+      "loss": 1.4178,
+      "step": 343
+    },
+    {
+      "epoch": 0.05407848454479357,
+      "grad_norm": 0.14452503621578217,
+      "learning_rate": 4.991538929148394e-05,
+      "loss": 1.4248,
+      "step": 344
+    },
+    {
+      "epoch": 0.05423568944172611,
+      "grad_norm": 0.1352955400943756,
+      "learning_rate": 4.991488064503674e-05,
+      "loss": 1.4304,
+      "step": 345
+    },
+    {
+      "epoch": 0.05439289433865865,
+      "grad_norm": 0.14846469461917877,
+      "learning_rate": 4.991437047688343e-05,
+      "loss": 1.4784,
+      "step": 346
+    },
+    {
+      "epoch": 0.05455009923559119,
+      "grad_norm": 0.12475849688053131,
+      "learning_rate": 4.9913858787055156e-05,
+      "loss": 1.4131,
+      "step": 347
+    },
+    {
+      "epoch": 0.054707304132523726,
+      "grad_norm": 0.13835409283638,
+      "learning_rate": 4.991334557558318e-05,
+      "loss": 1.4913,
+      "step": 348
+    },
+    {
+      "epoch": 0.05486450902945627,
+      "grad_norm": 0.13921529054641724,
+      "learning_rate": 4.991283084249885e-05,
+      "loss": 1.3713,
+      "step": 349
+    },
+    {
+      "epoch": 0.055021713926388806,
+      "grad_norm": 0.13188250362873077,
+      "learning_rate": 4.9912314587833586e-05,
+      "loss": 1.3608,
+      "step": 350
+    },
+    {
+      "epoch": 0.05517891882332135,
+      "grad_norm": 0.12457428872585297,
+      "learning_rate": 4.991179681161895e-05,
+      "loss": 1.4427,
+      "step": 351
+    },
+    {
+      "epoch": 0.055336123720253885,
+      "grad_norm": 0.12452542781829834,
+      "learning_rate": 4.9911277513886535e-05,
+      "loss": 1.4179,
+      "step": 352
+    },
+    {
+      "epoch": 0.05549332861718643,
+      "grad_norm": 0.14799195528030396,
+      "learning_rate": 4.9910756694668074e-05,
+      "loss": 1.4532,
+      "step": 353
+    },
+    {
+      "epoch": 0.055650533514118965,
+      "grad_norm": 0.13485541939735413,
+      "learning_rate": 4.991023435399538e-05,
+      "loss": 1.4114,
+      "step": 354
+    },
+    {
+      "epoch": 0.0558077384110515,
+      "grad_norm": 0.1422443389892578,
+      "learning_rate": 4.990971049190034e-05,
+      "loss": 1.377,
+      "step": 355
+    },
+    {
+      "epoch": 0.055964943307984044,
+      "grad_norm": 0.12994804978370667,
+      "learning_rate": 4.990918510841496e-05,
+      "loss": 1.4474,
+      "step": 356
+    },
+    {
+      "epoch": 0.05612214820491658,
+      "grad_norm": 0.1429785192012787,
+      "learning_rate": 4.990865820357133e-05,
+      "loss": 1.4435,
+      "step": 357
+    },
+    {
+      "epoch": 0.056279353101849124,
+      "grad_norm": 0.12979790568351746,
+      "learning_rate": 4.9908129777401625e-05,
+      "loss": 1.4039,
+      "step": 358
+    },
+    {
+      "epoch": 0.05643655799878166,
+      "grad_norm": 0.1332644671201706,
+      "learning_rate": 4.990759982993812e-05,
+      "loss": 1.4377,
+      "step": 359
+    },
+    {
+      "epoch": 0.0565937628957142,
+      "grad_norm": 0.13796579837799072,
+      "learning_rate": 4.99070683612132e-05,
+      "loss": 1.3951,
+      "step": 360
+    },
+    {
+      "epoch": 0.05675096779264674,
+      "grad_norm": 0.14315246045589447,
+      "learning_rate": 4.9906535371259294e-05,
+      "loss": 1.4042,
+      "step": 361
+    },
+    {
+      "epoch": 0.05690817268957928,
+      "grad_norm": 0.1463768631219864,
+      "learning_rate": 4.9906000860108974e-05,
+      "loss": 1.461,
+      "step": 362
+    },
+    {
+      "epoch": 0.05706537758651182,
+      "grad_norm": 0.14041170477867126,
+      "learning_rate": 4.9905464827794884e-05,
+      "loss": 1.4147,
+      "step": 363
+    },
+    {
+      "epoch": 0.05722258248344436,
+      "grad_norm": 0.19242697954177856,
+      "learning_rate": 4.990492727434976e-05,
+      "loss": 1.3435,
+      "step": 364
+    },
+    {
+      "epoch": 0.0573797873803769,
+      "grad_norm": 0.1556611955165863,
+      "learning_rate": 4.990438819980644e-05,
+      "loss": 1.4075,
+      "step": 365
+    },
+    {
+      "epoch": 0.05753699227730944,
+      "grad_norm": 0.13157570362091064,
+      "learning_rate": 4.990384760419784e-05,
+      "loss": 1.3334,
+      "step": 366
+    },
+    {
+      "epoch": 0.05769419717424198,
+      "grad_norm": 0.17953743040561676,
+      "learning_rate": 4.990330548755698e-05,
+      "loss": 1.4609,
+      "step": 367
+    },
+    {
+      "epoch": 0.057851402071174514,
+      "grad_norm": 0.14179491996765137,
+      "learning_rate": 4.990276184991697e-05,
+      "loss": 1.4344,
+      "step": 368
+    },
+    {
+      "epoch": 0.05800860696810706,
+      "grad_norm": 0.16522593796253204,
+      "learning_rate": 4.9902216691311024e-05,
+      "loss": 1.3794,
+      "step": 369
+    },
+    {
+      "epoch": 0.05816581186503959,
+      "grad_norm": 0.12736016511917114,
+      "learning_rate": 4.9901670011772425e-05,
+      "loss": 1.4167,
+      "step": 370
+    },
+    {
+      "epoch": 0.058323016761972137,
+      "grad_norm": 0.15869787335395813,
+      "learning_rate": 4.990112181133456e-05,
+      "loss": 1.4293,
+      "step": 371
+    },
+    {
+      "epoch": 0.05848022165890467,
+      "grad_norm": 0.14410504698753357,
+      "learning_rate": 4.990057209003093e-05,
+      "loss": 1.4357,
+      "step": 372
+    },
+    {
+      "epoch": 0.058637426555837216,
+      "grad_norm": 0.1567080020904541,
+      "learning_rate": 4.9900020847895086e-05,
+      "loss": 1.4146,
+      "step": 373
+    },
+    {
+      "epoch": 0.05879463145276975,
+      "grad_norm": 0.1430107057094574,
+      "learning_rate": 4.989946808496071e-05,
+      "loss": 1.3415,
+      "step": 374
+    },
+    {
+      "epoch": 0.058951836349702295,
+      "grad_norm": 0.146332785487175,
+      "learning_rate": 4.989891380126156e-05,
+      "loss": 1.496,
+      "step": 375
+    },
+    {
+      "epoch": 0.05910904124663483,
+      "grad_norm": 0.13674487173557281,
+      "learning_rate": 4.989835799683149e-05,
+      "loss": 1.3611,
+      "step": 376
+    },
+    {
+      "epoch": 0.059266246143567375,
+      "grad_norm": 0.1321984827518463,
+      "learning_rate": 4.989780067170444e-05,
+      "loss": 1.4695,
+      "step": 377
+    },
+    {
+      "epoch": 0.05942345104049991,
+      "grad_norm": 0.1535942554473877,
+      "learning_rate": 4.9897241825914464e-05,
+      "loss": 1.3564,
+      "step": 378
+    },
+    {
+      "epoch": 0.059580655937432454,
+      "grad_norm": 0.1538037806749344,
+      "learning_rate": 4.989668145949568e-05,
+      "loss": 1.3502,
+      "step": 379
+    },
+    {
+      "epoch": 0.05973786083436499,
+      "grad_norm": 0.15744829177856445,
+      "learning_rate": 4.989611957248232e-05,
+      "loss": 1.4318,
+      "step": 380
+    },
+    {
+      "epoch": 0.05989506573129753,
+      "grad_norm": 0.17178332805633545,
+      "learning_rate": 4.98955561649087e-05,
+      "loss": 1.4643,
+      "step": 381
+    },
+    {
+      "epoch": 0.06005227062823007,
+      "grad_norm": 0.15913072228431702,
+      "learning_rate": 4.989499123680923e-05,
+      "loss": 1.487,
+      "step": 382
+    },
+    {
+      "epoch": 0.060209475525162606,
+      "grad_norm": 0.15134060382843018,
+      "learning_rate": 4.9894424788218415e-05,
+      "loss": 1.4705,
+      "step": 383
+    },
+    {
+      "epoch": 0.06036668042209515,
+      "grad_norm": 0.13704389333724976,
+      "learning_rate": 4.989385681917085e-05,
+      "loss": 1.4756,
+      "step": 384
+    },
+    {
+      "epoch": 0.060523885319027686,
+      "grad_norm": 0.14025503396987915,
+      "learning_rate": 4.989328732970122e-05,
+      "loss": 1.443,
+      "step": 385
+    },
+    {
+      "epoch": 0.06068109021596023,
+      "grad_norm": 0.1822325438261032,
+      "learning_rate": 4.9892716319844325e-05,
+      "loss": 1.3996,
+      "step": 386
+    },
+    {
+      "epoch": 0.060838295112892765,
+      "grad_norm": 0.15639656782150269,
+      "learning_rate": 4.989214378963502e-05,
+      "loss": 1.3656,
+      "step": 387
+    },
+    {
+      "epoch": 0.06099550000982531,
+      "grad_norm": 0.15097728371620178,
+      "learning_rate": 4.989156973910828e-05,
+      "loss": 1.4055,
+      "step": 388
+    },
+    {
+      "epoch": 0.061152704906757845,
+      "grad_norm": 0.18977142870426178,
+      "learning_rate": 4.989099416829917e-05,
+      "loss": 1.4472,
+      "step": 389
+    },
+    {
+      "epoch": 0.06130990980369039,
+      "grad_norm": 0.1596304178237915,
+      "learning_rate": 4.989041707724284e-05,
+      "loss": 1.4373,
+      "step": 390
+    },
+    {
+      "epoch": 0.061467114700622924,
+      "grad_norm": 0.171820729970932,
+      "learning_rate": 4.988983846597454e-05,
+      "loss": 1.468,
+      "step": 391
+    },
+    {
+      "epoch": 0.06162431959755547,
+      "grad_norm": 0.14266176521778107,
+      "learning_rate": 4.98892583345296e-05,
+      "loss": 1.4037,
+      "step": 392
+    },
+    {
+      "epoch": 0.061781524494488003,
+      "grad_norm": 0.13375528156757355,
+      "learning_rate": 4.988867668294346e-05,
+      "loss": 1.437,
+      "step": 393
+    },
+    {
+      "epoch": 0.06193872939142054,
+      "grad_norm": 0.13332228362560272,
+      "learning_rate": 4.988809351125165e-05,
+      "loss": 1.3892,
+      "step": 394
+    },
+    {
+      "epoch": 0.06209593428835308,
+      "grad_norm": 0.17180980741977692,
+      "learning_rate": 4.988750881948977e-05,
+      "loss": 1.3494,
+      "step": 395
+    },
+    {
+      "epoch": 0.06225313918528562,
+      "grad_norm": 0.1419111043214798,
+      "learning_rate": 4.988692260769355e-05,
+      "loss": 1.3748,
+      "step": 396
+    },
+    {
+      "epoch": 0.06241034408221816,
+      "grad_norm": 0.17256620526313782,
+      "learning_rate": 4.9886334875898776e-05,
+      "loss": 1.3549,
+      "step": 397
+    },
+    {
+      "epoch": 0.0625675489791507,
+      "grad_norm": 0.2243422418832779,
+      "learning_rate": 4.988574562414137e-05,
+      "loss": 1.4465,
+      "step": 398
+    },
+    {
+      "epoch": 0.06272475387608324,
+      "grad_norm": 0.15700723230838776,
+      "learning_rate": 4.9885154852457294e-05,
+      "loss": 1.4477,
+      "step": 399
+    },
+    {
+      "epoch": 0.06288195877301578,
+      "grad_norm": 0.14497259259223938,
+      "learning_rate": 4.988456256088264e-05,
+      "loss": 1.3861,
+      "step": 400
+    },
+    {
+      "epoch": 0.06303916366994831,
+      "grad_norm": 0.14747034013271332,
+      "learning_rate": 4.988396874945359e-05,
+      "loss": 1.4206,
+      "step": 401
+    },
+    {
+      "epoch": 0.06319636856688086,
+      "grad_norm": 0.17671054601669312,
+      "learning_rate": 4.98833734182064e-05,
+      "loss": 1.2475,
+      "step": 402
+    },
+    {
+      "epoch": 0.0633535734638134,
+      "grad_norm": 0.16974316537380219,
+      "learning_rate": 4.9882776567177446e-05,
+      "loss": 1.4955,
+      "step": 403
+    },
+    {
+      "epoch": 0.06351077836074594,
+      "grad_norm": 0.15419775247573853,
+      "learning_rate": 4.988217819640317e-05,
+      "loss": 1.4209,
+      "step": 404
+    },
+    {
+      "epoch": 0.06366798325767847,
+      "grad_norm": 0.13987664878368378,
+      "learning_rate": 4.988157830592012e-05,
+      "loss": 1.456,
+      "step": 405
+    },
+    {
+      "epoch": 0.06382518815461101,
+      "grad_norm": 0.24560455977916718,
+      "learning_rate": 4.988097689576493e-05,
+      "loss": 1.3567,
+      "step": 406
+    },
+    {
+      "epoch": 0.06398239305154356,
+      "grad_norm": 0.13870076835155487,
+      "learning_rate": 4.9880373965974334e-05,
+      "loss": 1.3752,
+      "step": 407
+    },
+    {
+      "epoch": 0.0641395979484761,
+      "grad_norm": 0.16167718172073364,
+      "learning_rate": 4.987976951658517e-05,
+      "loss": 1.4766,
+      "step": 408
+    },
+    {
+      "epoch": 0.06429680284540863,
+      "grad_norm": 0.1700398474931717,
+      "learning_rate": 4.9879163547634346e-05,
+      "loss": 1.427,
+      "step": 409
+    },
+    {
+      "epoch": 0.06445400774234117,
+      "grad_norm": 0.15502458810806274,
+      "learning_rate": 4.987855605915887e-05,
+      "loss": 1.3965,
+      "step": 410
+    },
+    {
+      "epoch": 0.06461121263927372,
+      "grad_norm": 0.14834032952785492,
+      "learning_rate": 4.987794705119584e-05,
+      "loss": 1.4399,
+      "step": 411
+    },
+    {
+      "epoch": 0.06476841753620625,
+      "grad_norm": 0.22443649172782898,
+      "learning_rate": 4.987733652378246e-05,
+      "loss": 1.3736,
+      "step": 412
+    },
+    {
+      "epoch": 0.06492562243313879,
+      "grad_norm": 0.14396560192108154,
+      "learning_rate": 4.9876724476956015e-05,
+      "loss": 1.4648,
+      "step": 413
+    },
+    {
+      "epoch": 0.06508282733007133,
+      "grad_norm": 0.15352006256580353,
+      "learning_rate": 4.987611091075389e-05,
+      "loss": 1.4988,
+      "step": 414
+    },
+    {
+      "epoch": 0.06524003222700388,
+      "grad_norm": 0.13210074603557587,
+      "learning_rate": 4.987549582521356e-05,
+      "loss": 1.3705,
+      "step": 415
+    },
+    {
+      "epoch": 0.06539723712393641,
+      "grad_norm": 0.16056782007217407,
+      "learning_rate": 4.98748792203726e-05,
+      "loss": 1.3388,
+      "step": 416
+    },
+    {
+      "epoch": 0.06555444202086895,
+      "grad_norm": 0.18992343544960022,
+      "learning_rate": 4.9874261096268647e-05,
+      "loss": 1.3842,
+      "step": 417
+    },
+    {
+      "epoch": 0.06571164691780149,
+      "grad_norm": 0.1789916455745697,
+      "learning_rate": 4.9873641452939466e-05,
+      "loss": 1.3622,
+      "step": 418
+    },
+    {
+      "epoch": 0.06586885181473402,
+      "grad_norm": 0.21043789386749268,
+      "learning_rate": 4.9873020290422915e-05,
+      "loss": 1.3477,
+      "step": 419
+    },
+    {
+      "epoch": 0.06602605671166657,
+      "grad_norm": 0.15355254709720612,
+      "learning_rate": 4.987239760875691e-05,
+      "loss": 1.3643,
+      "step": 420
+    },
+    {
+      "epoch": 0.06618326160859911,
+      "grad_norm": 0.1433190107345581,
+      "learning_rate": 4.9871773407979496e-05,
+      "loss": 1.3753,
+      "step": 421
+    },
+    {
+      "epoch": 0.06634046650553165,
+      "grad_norm": 0.17479249835014343,
+      "learning_rate": 4.987114768812879e-05,
+      "loss": 1.3809,
+      "step": 422
+    },
+    {
+      "epoch": 0.06649767140246418,
+      "grad_norm": 0.186944842338562,
+      "learning_rate": 4.987052044924302e-05,
+      "loss": 1.3616,
+      "step": 423
+    },
+    {
+      "epoch": 0.06665487629939673,
+      "grad_norm": 0.15202952921390533,
+      "learning_rate": 4.986989169136048e-05,
+      "loss": 1.4479,
+      "step": 424
+    },
+    {
+      "epoch": 0.06681208119632927,
+      "grad_norm": 0.16295532882213593,
+      "learning_rate": 4.9869261414519575e-05,
+      "loss": 1.3713,
+      "step": 425
+    },
+    {
+      "epoch": 0.0669692860932618,
+      "grad_norm": 0.19577625393867493,
+      "learning_rate": 4.986862961875881e-05,
+      "loss": 1.4199,
+      "step": 426
+    },
+    {
+      "epoch": 0.06712649099019434,
+      "grad_norm": 0.22768542170524597,
+      "learning_rate": 4.986799630411677e-05,
+      "loss": 1.3529,
+      "step": 427
+    },
+    {
+      "epoch": 0.06728369588712689,
+      "grad_norm": 0.25184011459350586,
+      "learning_rate": 4.986736147063212e-05,
+      "loss": 1.3944,
+      "step": 428
+    },
+    {
+      "epoch": 0.06744090078405943,
+      "grad_norm": 0.15565118193626404,
+      "learning_rate": 4.986672511834366e-05,
+      "loss": 1.4505,
+      "step": 429
+    },
+    {
+      "epoch": 0.06759810568099196,
+      "grad_norm": 0.16559922695159912,
+      "learning_rate": 4.986608724729024e-05,
+      "loss": 1.3742,
+      "step": 430
+    },
+    {
+      "epoch": 0.0677553105779245,
+      "grad_norm": 0.14826242625713348,
+      "learning_rate": 4.986544785751081e-05,
+      "loss": 1.4008,
+      "step": 431
+    },
+    {
+      "epoch": 0.06791251547485704,
+      "grad_norm": 0.16543184220790863,
+      "learning_rate": 4.986480694904444e-05,
+      "loss": 1.3433,
+      "step": 432
+    },
+    {
+      "epoch": 0.06806972037178959,
+      "grad_norm": 0.15332931280136108,
+      "learning_rate": 4.986416452193027e-05,
+      "loss": 1.4459,
+      "step": 433
+    },
+    {
+      "epoch": 0.06822692526872212,
+      "grad_norm": 0.18880733847618103,
+      "learning_rate": 4.986352057620752e-05,
+      "loss": 1.3902,
+      "step": 434
+    },
+    {
+      "epoch": 0.06838413016565466,
+      "grad_norm": 0.1513829231262207,
+      "learning_rate": 4.986287511191554e-05,
+      "loss": 1.3485,
+      "step": 435
+    },
+    {
+      "epoch": 0.0685413350625872,
+      "grad_norm": 0.15241704881191254,
+      "learning_rate": 4.9862228129093745e-05,
+      "loss": 1.3051,
+      "step": 436
+    },
+    {
+      "epoch": 0.06869853995951974,
+      "grad_norm": 0.1956702321767807,
+      "learning_rate": 4.986157962778165e-05,
+      "loss": 1.4647,
+      "step": 437
+    },
+    {
+      "epoch": 0.06885574485645228,
+      "grad_norm": 0.2027936428785324,
+      "learning_rate": 4.9860929608018866e-05,
+      "loss": 1.3602,
+      "step": 438
+    },
+    {
+      "epoch": 0.06901294975338482,
+      "grad_norm": 0.1623186320066452,
+      "learning_rate": 4.986027806984509e-05,
+      "loss": 1.4154,
+      "step": 439
+    },
+    {
+      "epoch": 0.06917015465031735,
+      "grad_norm": 0.16111283004283905,
+      "learning_rate": 4.985962501330011e-05,
+      "loss": 1.4311,
+      "step": 440
+    },
+    {
+      "epoch": 0.0693273595472499,
+      "grad_norm": 0.16754299402236938,
+      "learning_rate": 4.985897043842382e-05,
+      "loss": 1.349,
+      "step": 441
+    },
+    {
+      "epoch": 0.06948456444418244,
+      "grad_norm": 0.1766330897808075,
+      "learning_rate": 4.985831434525621e-05,
+      "loss": 1.3714,
+      "step": 442
+    },
+    {
+      "epoch": 0.06964176934111498,
+      "grad_norm": 0.1742810308933258,
+      "learning_rate": 4.985765673383733e-05,
+      "loss": 1.4161,
+      "step": 443
+    },
+    {
+      "epoch": 0.06979897423804751,
+      "grad_norm": 0.17025281488895416,
+      "learning_rate": 4.985699760420736e-05,
+      "loss": 1.3925,
+      "step": 444
+    },
+    {
+      "epoch": 0.06995617913498005,
+      "grad_norm": 0.19201375544071198,
+      "learning_rate": 4.985633695640655e-05,
+      "loss": 1.4158,
+      "step": 445
+    },
+    {
+      "epoch": 0.0701133840319126,
+      "grad_norm": 0.1636267751455307,
+      "learning_rate": 4.985567479047524e-05,
+      "loss": 1.4071,
+      "step": 446
+    },
+    {
+      "epoch": 0.07027058892884513,
+      "grad_norm": 0.19676333665847778,
+      "learning_rate": 4.9855011106453894e-05,
+      "loss": 1.3449,
+      "step": 447
+    },
+    {
+      "epoch": 0.07042779382577767,
+      "grad_norm": 0.17712907493114471,
+      "learning_rate": 4.985434590438303e-05,
+      "loss": 1.3421,
+      "step": 448
+    },
+    {
+      "epoch": 0.07058499872271021,
+      "grad_norm": 0.18515101075172424,
+      "learning_rate": 4.985367918430329e-05,
+      "loss": 1.4051,
+      "step": 449
+    },
+    {
+      "epoch": 0.07074220361964276,
+      "grad_norm": 0.17168915271759033,
+      "learning_rate": 4.985301094625538e-05,
+      "loss": 1.3093,
+      "step": 450
+    },
+    {
+      "epoch": 0.0708994085165753,
+      "grad_norm": 0.1891397386789322,
+      "learning_rate": 4.9852341190280127e-05,
+      "loss": 1.3075,
+      "step": 451
+    },
+    {
+      "epoch": 0.07105661341350783,
+      "grad_norm": 0.17731457948684692,
+      "learning_rate": 4.985166991641843e-05,
+      "loss": 1.3986,
+      "step": 452
+    },
+    {
+      "epoch": 0.07121381831044037,
+      "grad_norm": 0.18817296624183655,
+      "learning_rate": 4.985099712471129e-05,
+      "loss": 1.3531,
+      "step": 453
+    },
+    {
+      "epoch": 0.07137102320737292,
+      "grad_norm": 0.1782791018486023,
+      "learning_rate": 4.9850322815199795e-05,
+      "loss": 1.4064,
+      "step": 454
+    },
+    {
+      "epoch": 0.07152822810430545,
+      "grad_norm": 0.18053874373435974,
+      "learning_rate": 4.984964698792514e-05,
+      "loss": 1.4607,
+      "step": 455
+    },
+    {
+      "epoch": 0.07168543300123799,
+      "grad_norm": 0.286338746547699,
+      "learning_rate": 4.984896964292858e-05,
+      "loss": 1.3036,
+      "step": 456
+    },
+    {
+      "epoch": 0.07184263789817052,
+      "grad_norm": 0.2560707926750183,
+      "learning_rate": 4.98482907802515e-05,
+      "loss": 1.3428,
+      "step": 457
+    },
+    {
+      "epoch": 0.07199984279510306,
+      "grad_norm": 0.19296897947788239,
+      "learning_rate": 4.984761039993537e-05,
+      "loss": 1.3502,
+      "step": 458
+    },
+    {
+      "epoch": 0.07215704769203561,
+      "grad_norm": 0.19685949385166168,
+      "learning_rate": 4.9846928502021725e-05,
+      "loss": 1.4015,
+      "step": 459
+    },
+    {
+      "epoch": 0.07231425258896815,
+      "grad_norm": 0.1548481583595276,
+      "learning_rate": 4.984624508655223e-05,
+      "loss": 1.3698,
+      "step": 460
+    },
+    {
+      "epoch": 0.07247145748590068,
+      "grad_norm": 0.16076034307479858,
+      "learning_rate": 4.984556015356862e-05,
+      "loss": 1.3627,
+      "step": 461
+    },
+    {
+      "epoch": 0.07262866238283322,
+      "grad_norm": 0.18571603298187256,
+      "learning_rate": 4.9844873703112726e-05,
+      "loss": 1.3506,
+      "step": 462
+    },
+    {
+      "epoch": 0.07278586727976577,
+      "grad_norm": 0.1540035605430603,
+      "learning_rate": 4.984418573522648e-05,
+      "loss": 1.4483,
+      "step": 463
+    },
+    {
+      "epoch": 0.0729430721766983,
+      "grad_norm": 0.1730145364999771,
+      "learning_rate": 4.984349624995188e-05,
+      "loss": 1.3678,
+      "step": 464
+    },
+    {
+      "epoch": 0.07310027707363084,
+      "grad_norm": 0.26254212856292725,
+      "learning_rate": 4.984280524733107e-05,
+      "loss": 1.401,
+      "step": 465
+    },
+    {
+      "epoch": 0.07325748197056338,
+      "grad_norm": 0.2079063057899475,
+      "learning_rate": 4.984211272740623e-05,
+      "loss": 1.3655,
+      "step": 466
+    },
+    {
+      "epoch": 0.07341468686749593,
+      "grad_norm": 0.21711499989032745,
+      "learning_rate": 4.9841418690219653e-05,
+      "loss": 1.4011,
+      "step": 467
+    },
+    {
+      "epoch": 0.07357189176442847,
+      "grad_norm": 0.18226252496242523,
+      "learning_rate": 4.984072313581375e-05,
+      "loss": 1.4213,
+      "step": 468
+    },
+    {
+      "epoch": 0.073729096661361,
+      "grad_norm": 0.1463780552148819,
+      "learning_rate": 4.9840026064230984e-05,
+      "loss": 1.4519,
+      "step": 469
+    },
+    {
+      "epoch": 0.07388630155829354,
+      "grad_norm": 0.18232892453670502,
+      "learning_rate": 4.983932747551394e-05,
+      "loss": 1.3657,
+      "step": 470
+    },
+    {
+      "epoch": 0.07404350645522607,
+      "grad_norm": 0.19644559919834137,
+      "learning_rate": 4.9838627369705285e-05,
+      "loss": 1.3988,
+      "step": 471
+    },
+    {
+      "epoch": 0.07420071135215862,
+      "grad_norm": 0.16292576491832733,
+      "learning_rate": 4.983792574684776e-05,
+      "loss": 1.4369,
+      "step": 472
+    },
+    {
+      "epoch": 0.07435791624909116,
+      "grad_norm": 0.2244543433189392,
+      "learning_rate": 4.983722260698425e-05,
+      "loss": 1.4269,
+      "step": 473
+    },
+    {
+      "epoch": 0.0745151211460237,
+      "grad_norm": 0.2582489848136902,
+      "learning_rate": 4.9836517950157666e-05,
+      "loss": 1.3986,
+      "step": 474
+    },
+    {
+      "epoch": 0.07467232604295623,
+      "grad_norm": 0.15564194321632385,
+      "learning_rate": 4.983581177641108e-05,
+      "loss": 1.3871,
+      "step": 475
+    },
+    {
+      "epoch": 0.07482953093988878,
+      "grad_norm": 0.2301008552312851,
+      "learning_rate": 4.9835104085787596e-05,
+      "loss": 1.3572,
+      "step": 476
+    },
+    {
+      "epoch": 0.07498673583682132,
+      "grad_norm": 0.21603424847126007,
+      "learning_rate": 4.9834394878330444e-05,
+      "loss": 1.3803,
+      "step": 477
+    },
+    {
+      "epoch": 0.07514394073375386,
+      "grad_norm": 0.16744717955589294,
+      "learning_rate": 4.9833684154082937e-05,
+      "loss": 1.4233,
+      "step": 478
+    },
+    {
+      "epoch": 0.07530114563068639,
+      "grad_norm": 0.23016415536403656,
+      "learning_rate": 4.98329719130885e-05,
+      "loss": 1.3962,
+      "step": 479
+    },
+    {
+      "epoch": 0.07545835052761894,
+      "grad_norm": 0.19687114655971527,
+      "learning_rate": 4.983225815539061e-05,
+      "loss": 1.3667,
+      "step": 480
+    },
+    {
+      "epoch": 0.07545835052761894,
+      "eval_loss": 1.3748993873596191,
+      "eval_runtime": 2315.5952,
+      "eval_samples_per_second": 3.998,
+      "eval_steps_per_second": 1.999,
+      "step": 480
+    },
+    {
+      "epoch": 0.07561555542455148,
+      "grad_norm": 0.1833205670118332,
+      "learning_rate": 4.9831542881032884e-05,
+      "loss": 1.4365,
+      "step": 481
+    },
+    {
+      "epoch": 0.07577276032148401,
+      "grad_norm": 0.17124423384666443,
+      "learning_rate": 4.983082609005899e-05,
+      "loss": 1.3641,
+      "step": 482
+    },
+    {
+      "epoch": 0.07592996521841655,
+      "grad_norm": 0.17352670431137085,
+      "learning_rate": 4.9830107782512715e-05,
+      "loss": 1.3415,
+      "step": 483
+    },
+    {
+      "epoch": 0.07608717011534909,
+      "grad_norm": 0.20768220722675323,
+      "learning_rate": 4.982938795843793e-05,
+      "loss": 1.3261,
+      "step": 484
+    },
+    {
+      "epoch": 0.07624437501228164,
+      "grad_norm": 0.21459853649139404,
+      "learning_rate": 4.982866661787859e-05,
+      "loss": 1.4185,
+      "step": 485
+    },
+    {
+      "epoch": 0.07640157990921417,
+      "grad_norm": 0.26912233233451843,
+      "learning_rate": 4.982794376087877e-05,
+      "loss": 1.3941,
+      "step": 486
+    },
+    {
+      "epoch": 0.07655878480614671,
+      "grad_norm": 0.28497114777565,
+      "learning_rate": 4.982721938748261e-05,
+      "loss": 1.3201,
+      "step": 487
+    },
+    {
+      "epoch": 0.07671598970307925,
+      "grad_norm": 0.15378472208976746,
+      "learning_rate": 4.982649349773435e-05,
+      "loss": 1.3615,
+      "step": 488
+    },
+    {
+      "epoch": 0.0768731946000118,
+      "grad_norm": 0.16169893741607666,
+      "learning_rate": 4.982576609167831e-05,
+      "loss": 1.3342,
+      "step": 489
+    },
+    {
+      "epoch": 0.07703039949694433,
+      "grad_norm": 0.24693650007247925,
+      "learning_rate": 4.982503716935896e-05,
+      "loss": 1.3788,
+      "step": 490
+    },
+    {
+      "epoch": 0.07718760439387687,
+      "grad_norm": 0.1769181787967682,
+      "learning_rate": 4.982430673082077e-05,
+      "loss": 1.3664,
+      "step": 491
+    },
+    {
+      "epoch": 0.0773448092908094,
+      "grad_norm": 0.26325106620788574,
+      "learning_rate": 4.982357477610839e-05,
+      "loss": 1.3173,
+      "step": 492
+    },
+    {
+      "epoch": 0.07750201418774195,
+      "grad_norm": 0.2063319832086563,
+      "learning_rate": 4.9822841305266506e-05,
+      "loss": 1.4125,
+      "step": 493
+    },
+    {
+      "epoch": 0.07765921908467449,
+      "grad_norm": 0.29141879081726074,
+      "learning_rate": 4.982210631833992e-05,
+      "loss": 1.3596,
+      "step": 494
+    },
+    {
+      "epoch": 0.07781642398160703,
+      "grad_norm": 0.18967591226100922,
+      "learning_rate": 4.982136981537352e-05,
+      "loss": 1.4128,
+      "step": 495
+    },
+    {
+      "epoch": 0.07797362887853956,
+      "grad_norm": 0.2291795313358307,
+      "learning_rate": 4.9820631796412287e-05,
+      "loss": 1.3772,
+      "step": 496
+    },
+    {
+      "epoch": 0.0781308337754721,
+      "grad_norm": 0.200834721326828,
+      "learning_rate": 4.98198922615013e-05,
+      "loss": 1.369,
+      "step": 497
+    },
+    {
+      "epoch": 0.07828803867240465,
+      "grad_norm": 0.22960609197616577,
+      "learning_rate": 4.9819151210685736e-05,
+      "loss": 1.3979,
+      "step": 498
+    },
+    {
+      "epoch": 0.07844524356933719,
+      "grad_norm": 0.17247427999973297,
+      "learning_rate": 4.981840864401084e-05,
+      "loss": 1.3927,
+      "step": 499
+    },
+    {
+      "epoch": 0.07860244846626972,
+      "grad_norm": 0.2623608112335205,
+      "learning_rate": 4.981766456152198e-05,
+      "loss": 1.3919,
+      "step": 500
+    },
+    {
+      "epoch": 0.07875965336320226,
+      "grad_norm": 0.19911788403987885,
+      "learning_rate": 4.981691896326459e-05,
+      "loss": 1.3925,
+      "step": 501
+    },
+    {
+      "epoch": 0.07891685826013481,
+      "grad_norm": 0.24869734048843384,
+      "learning_rate": 4.9816171849284205e-05,
+      "loss": 1.3562,
+      "step": 502
+    },
+    {
+      "epoch": 0.07907406315706735,
+      "grad_norm": 0.31372350454330444,
+      "learning_rate": 4.981542321962647e-05,
+      "loss": 1.3211,
+      "step": 503
+    },
+    {
+      "epoch": 0.07923126805399988,
+      "grad_norm": 0.21760910749435425,
+      "learning_rate": 4.981467307433709e-05,
+      "loss": 1.3042,
+      "step": 504
+    },
+    {
+      "epoch": 0.07938847295093242,
+      "grad_norm": 0.2469843477010727,
+      "learning_rate": 4.9813921413461906e-05,
+      "loss": 1.2831,
+      "step": 505
+    },
+    {
+      "epoch": 0.07954567784786497,
+      "grad_norm": 0.24319148063659668,
+      "learning_rate": 4.981316823704681e-05,
+      "loss": 1.2703,
+      "step": 506
+    },
+    {
+      "epoch": 0.0797028827447975,
+      "grad_norm": 0.19718031585216522,
+      "learning_rate": 4.98124135451378e-05,
+      "loss": 1.3258,
+      "step": 507
+    },
+    {
+      "epoch": 0.07986008764173004,
+      "grad_norm": 0.17459236085414886,
+      "learning_rate": 4.981165733778098e-05,
+      "loss": 1.4248,
+      "step": 508
+    },
+    {
+      "epoch": 0.08001729253866258,
+      "grad_norm": 0.17684616148471832,
+      "learning_rate": 4.981089961502253e-05,
+      "loss": 1.3939,
+      "step": 509
+    },
+    {
+      "epoch": 0.08017449743559511,
+      "grad_norm": 0.17499729990959167,
+      "learning_rate": 4.981014037690874e-05,
+      "loss": 1.4156,
+      "step": 510
+    },
+    {
+      "epoch": 0.08033170233252766,
+      "grad_norm": 0.1901170015335083,
+      "learning_rate": 4.9809379623485964e-05,
+      "loss": 1.4209,
+      "step": 511
+    },
+    {
+      "epoch": 0.0804889072294602,
+      "grad_norm": 0.18230682611465454,
+      "learning_rate": 4.980861735480067e-05,
+      "loss": 1.4607,
+      "step": 512
+    },
+    {
+      "epoch": 0.08064611212639274,
+      "grad_norm": 0.22843636572360992,
+      "learning_rate": 4.9807853570899427e-05,
+      "loss": 1.3671,
+      "step": 513
+    },
+    {
+      "epoch": 0.08080331702332527,
+      "grad_norm": 0.2288489192724228,
+      "learning_rate": 4.980708827182887e-05,
+      "loss": 1.3657,
+      "step": 514
+    },
+    {
+      "epoch": 0.08096052192025782,
+      "grad_norm": 0.19647593796253204,
+      "learning_rate": 4.980632145763575e-05,
+      "loss": 1.4079,
+      "step": 515
+    },
+    {
+      "epoch": 0.08111772681719036,
+      "grad_norm": 0.20980435609817505,
+      "learning_rate": 4.98055531283669e-05,
+      "loss": 1.3746,
+      "step": 516
+    },
+    {
+      "epoch": 0.0812749317141229,
+      "grad_norm": 0.19381123781204224,
+      "learning_rate": 4.980478328406923e-05,
+      "loss": 1.3986,
+      "step": 517
+    },
+    {
+      "epoch": 0.08143213661105543,
+      "grad_norm": 0.2224361151456833,
+      "learning_rate": 4.980401192478979e-05,
+      "loss": 1.3082,
+      "step": 518
+    },
+    {
+      "epoch": 0.08158934150798797,
+      "grad_norm": 0.20567384362220764,
+      "learning_rate": 4.9803239050575664e-05,
+      "loss": 1.4417,
+      "step": 519
+    },
+    {
+      "epoch": 0.08174654640492052,
+      "grad_norm": 0.22890503704547882,
+      "learning_rate": 4.9802464661474074e-05,
+      "loss": 1.3034,
+      "step": 520
+    },
+    {
+      "epoch": 0.08190375130185305,
+      "grad_norm": 0.23220910131931305,
+      "learning_rate": 4.9801688757532304e-05,
+      "loss": 1.3705,
+      "step": 521
+    },
+    {
+      "epoch": 0.08206095619878559,
+      "grad_norm": 0.29084959626197815,
+      "learning_rate": 4.980091133879775e-05,
+      "loss": 1.3246,
+      "step": 522
+    },
+    {
+      "epoch": 0.08221816109571813,
+      "grad_norm": 0.15776456892490387,
+      "learning_rate": 4.9800132405317895e-05,
+      "loss": 1.4311,
+      "step": 523
+    },
+    {
+      "epoch": 0.08237536599265068,
+      "grad_norm": 0.2636071443557739,
+      "learning_rate": 4.9799351957140314e-05,
+      "loss": 1.3265,
+      "step": 524
+    },
+    {
+      "epoch": 0.08253257088958321,
+      "grad_norm": 0.20042134821414948,
+      "learning_rate": 4.979856999431266e-05,
+      "loss": 1.3257,
+      "step": 525
+    },
+    {
+      "epoch": 0.08268977578651575,
+      "grad_norm": 0.24039289355278015,
+      "learning_rate": 4.9797786516882714e-05,
+      "loss": 1.3999,
+      "step": 526
+    },
+    {
+      "epoch": 0.08284698068344828,
+      "grad_norm": 0.16932524740695953,
+      "learning_rate": 4.9797001524898315e-05,
+      "loss": 1.4113,
+      "step": 527
+    },
+    {
+      "epoch": 0.08300418558038083,
+      "grad_norm": 0.2101370096206665,
+      "learning_rate": 4.97962150184074e-05,
+      "loss": 1.3973,
+      "step": 528
+    },
+    {
+      "epoch": 0.08316139047731337,
+      "grad_norm": 0.20983585715293884,
+      "learning_rate": 4.979542699745803e-05,
+      "loss": 1.3255,
+      "step": 529
+    },
+    {
+      "epoch": 0.08331859537424591,
+      "grad_norm": 0.20477800071239471,
+      "learning_rate": 4.97946374620983e-05,
+      "loss": 1.4349,
+      "step": 530
+    },
+    {
+      "epoch": 0.08347580027117844,
+      "grad_norm": 0.22637289762496948,
+      "learning_rate": 4.979384641237647e-05,
+      "loss": 1.3263,
+      "step": 531
+    },
+    {
+      "epoch": 0.08363300516811098,
+      "grad_norm": 0.20332221686840057,
+      "learning_rate": 4.9793053848340835e-05,
+      "loss": 1.3411,
+      "step": 532
+    },
+    {
+      "epoch": 0.08379021006504353,
+      "grad_norm": 0.22744616866111755,
+      "learning_rate": 4.979225977003979e-05,
+      "loss": 1.4042,
+      "step": 533
+    },
+    {
+      "epoch": 0.08394741496197607,
+      "grad_norm": 0.20091576874256134,
+      "learning_rate": 4.979146417752185e-05,
+      "loss": 1.3218,
+      "step": 534
+    },
+    {
+      "epoch": 0.0841046198589086,
+      "grad_norm": 0.2225920408964157,
+      "learning_rate": 4.9790667070835604e-05,
+      "loss": 1.4223,
+      "step": 535
+    },
+    {
+      "epoch": 0.08426182475584114,
+      "grad_norm": 0.20447570085525513,
+      "learning_rate": 4.9789868450029745e-05,
+      "loss": 1.3884,
+      "step": 536
+    },
+    {
+      "epoch": 0.08441902965277369,
+      "grad_norm": 0.22765719890594482,
+      "learning_rate": 4.9789068315153035e-05,
+      "loss": 1.3575,
+      "step": 537
+    },
+    {
+      "epoch": 0.08457623454970623,
+      "grad_norm": 0.18886259198188782,
+      "learning_rate": 4.9788266666254343e-05,
+      "loss": 1.2737,
+      "step": 538
+    },
+    {
+      "epoch": 0.08473343944663876,
+      "grad_norm": 0.26551586389541626,
+      "learning_rate": 4.978746350338264e-05,
+      "loss": 1.3867,
+      "step": 539
+    },
+    {
+      "epoch": 0.0848906443435713,
+      "grad_norm": 0.29268744587898254,
+      "learning_rate": 4.9786658826586975e-05,
+      "loss": 1.4266,
+      "step": 540
+    },
+    {
+      "epoch": 0.08504784924050385,
+      "grad_norm": 0.2537211775779724,
+      "learning_rate": 4.97858526359165e-05,
+      "loss": 1.3402,
+      "step": 541
+    },
+    {
+      "epoch": 0.08520505413743638,
+      "grad_norm": 0.20287925004959106,
+      "learning_rate": 4.978504493142045e-05,
+      "loss": 1.3148,
+      "step": 542
+    },
+    {
+      "epoch": 0.08536225903436892,
+      "grad_norm": 0.18584851920604706,
+      "learning_rate": 4.978423571314814e-05,
+      "loss": 1.3293,
+      "step": 543
+    },
+    {
+      "epoch": 0.08551946393130146,
+      "grad_norm": 0.1944153755903244,
+      "learning_rate": 4.978342498114903e-05,
+      "loss": 1.4084,
+      "step": 544
+    },
+    {
+      "epoch": 0.08567666882823399,
+      "grad_norm": 0.18139739334583282,
+      "learning_rate": 4.978261273547261e-05,
+      "loss": 1.2734,
+      "step": 545
+    },
+    {
+      "epoch": 0.08583387372516654,
+      "grad_norm": 0.20824116468429565,
+      "learning_rate": 4.97817989761685e-05,
+      "loss": 1.3346,
+      "step": 546
+    },
+    {
+      "epoch": 0.08599107862209908,
+      "grad_norm": 0.16180047392845154,
+      "learning_rate": 4.978098370328639e-05,
+      "loss": 1.4547,
+      "step": 547
+    },
+    {
+      "epoch": 0.08614828351903162,
+      "grad_norm": 0.17156392335891724,
+      "learning_rate": 4.978016691687609e-05,
+      "loss": 1.366,
+      "step": 548
+    },
+    {
+      "epoch": 0.08630548841596415,
+      "grad_norm": 0.17913401126861572,
+      "learning_rate": 4.977934861698746e-05,
+      "loss": 1.2771,
+      "step": 549
+    },
+    {
+      "epoch": 0.0864626933128967,
+      "grad_norm": 0.17393502593040466,
+      "learning_rate": 4.977852880367051e-05,
+      "loss": 1.3061,
+      "step": 550
+    },
+    {
+      "epoch": 0.08661989820982924,
+      "grad_norm": 0.21741637587547302,
+      "learning_rate": 4.97777074769753e-05,
+      "loss": 1.3232,
+      "step": 551
+    },
+    {
+      "epoch": 0.08677710310676177,
+      "grad_norm": 0.26123344898223877,
+      "learning_rate": 4.977688463695198e-05,
+      "loss": 1.2678,
+      "step": 552
+    },
+    {
+      "epoch": 0.08693430800369431,
+      "grad_norm": 0.2508600354194641,
+      "learning_rate": 4.9776060283650826e-05,
+      "loss": 1.4543,
+      "step": 553
+    },
+    {
+      "epoch": 0.08709151290062686,
+      "grad_norm": 0.18527132272720337,
+      "learning_rate": 4.977523441712217e-05,
+      "loss": 1.3359,
+      "step": 554
+    },
+    {
+      "epoch": 0.0872487177975594,
+      "grad_norm": 0.24495406448841095,
+      "learning_rate": 4.977440703741646e-05,
+      "loss": 1.2892,
+      "step": 555
+    },
+    {
+      "epoch": 0.08740592269449193,
+      "grad_norm": 0.22759339213371277,
+      "learning_rate": 4.9773578144584235e-05,
+      "loss": 1.2212,
+      "step": 556
+    },
+    {
+      "epoch": 0.08756312759142447,
+      "grad_norm": 0.1627693474292755,
+      "learning_rate": 4.977274773867611e-05,
+      "loss": 1.3461,
+      "step": 557
+    },
+    {
+      "epoch": 0.087720332488357,
+      "grad_norm": 0.2068985551595688,
+      "learning_rate": 4.9771915819742804e-05,
+      "loss": 1.3348,
+      "step": 558
+    },
+    {
+      "epoch": 0.08787753738528956,
+      "grad_norm": 0.19731195271015167,
+      "learning_rate": 4.9771082387835135e-05,
+      "loss": 1.3727,
+      "step": 559
+    },
+    {
+      "epoch": 0.08803474228222209,
+      "grad_norm": 0.26571184396743774,
+      "learning_rate": 4.977024744300399e-05,
+      "loss": 1.3911,
+      "step": 560
+    },
+    {
+      "epoch": 0.08819194717915463,
+      "grad_norm": 0.23141519725322723,
+      "learning_rate": 4.976941098530039e-05,
+      "loss": 1.3978,
+      "step": 561
+    },
+    {
+      "epoch": 0.08834915207608716,
+      "grad_norm": 0.2507224380970001,
+      "learning_rate": 4.97685730147754e-05,
+      "loss": 1.3017,
+      "step": 562
+    },
+    {
+      "epoch": 0.08850635697301971,
+      "grad_norm": 0.2453109323978424,
+      "learning_rate": 4.976773353148022e-05,
+      "loss": 1.2977,
+      "step": 563
+    },
+    {
+      "epoch": 0.08866356186995225,
+      "grad_norm": 0.2600953280925751,
+      "learning_rate": 4.9766892535466105e-05,
+      "loss": 1.4015,
+      "step": 564
+    },
+    {
+      "epoch": 0.08882076676688479,
+      "grad_norm": 0.19863371551036835,
+      "learning_rate": 4.9766050026784416e-05,
+      "loss": 1.3593,
+      "step": 565
+    },
+    {
+      "epoch": 0.08897797166381732,
+      "grad_norm": 0.2115338146686554,
+      "learning_rate": 4.976520600548663e-05,
+      "loss": 1.2928,
+      "step": 566
+    },
+    {
+      "epoch": 0.08913517656074987,
+      "grad_norm": 0.18994684517383575,
+      "learning_rate": 4.976436047162429e-05,
+      "loss": 1.3506,
+      "step": 567
+    },
+    {
+      "epoch": 0.08929238145768241,
+      "grad_norm": 0.22891771793365479,
+      "learning_rate": 4.976351342524903e-05,
+      "loss": 1.4449,
+      "step": 568
+    },
+    {
+      "epoch": 0.08944958635461495,
+      "grad_norm": 0.19313135743141174,
+      "learning_rate": 4.976266486641259e-05,
+      "loss": 1.2916,
+      "step": 569
+    },
+    {
+      "epoch": 0.08960679125154748,
+      "grad_norm": 0.17697346210479736,
+      "learning_rate": 4.976181479516679e-05,
+      "loss": 1.3696,
+      "step": 570
+    },
+    {
+      "epoch": 0.08976399614848002,
+      "grad_norm": 0.22902925312519073,
+      "learning_rate": 4.976096321156356e-05,
+      "loss": 1.3688,
+      "step": 571
+    },
+    {
+      "epoch": 0.08992120104541257,
+      "grad_norm": 0.25305554270744324,
+      "learning_rate": 4.97601101156549e-05,
+      "loss": 1.3057,
+      "step": 572
+    },
+    {
+      "epoch": 0.0900784059423451,
+      "grad_norm": 0.23255370557308197,
+      "learning_rate": 4.975925550749293e-05,
+      "loss": 1.3571,
+      "step": 573
+    },
+    {
+      "epoch": 0.09023561083927764,
+      "grad_norm": 0.25259101390838623,
+      "learning_rate": 4.9758399387129834e-05,
+      "loss": 1.3152,
+      "step": 574
+    },
+    {
+      "epoch": 0.09039281573621018,
+      "grad_norm": 0.26062390208244324,
+      "learning_rate": 4.97575417546179e-05,
+      "loss": 1.3042,
+      "step": 575
+    },
+    {
+      "epoch": 0.09055002063314273,
+      "grad_norm": 0.16536732017993927,
+      "learning_rate": 4.9756682610009515e-05,
+      "loss": 1.2797,
+      "step": 576
+    },
+    {
+      "epoch": 0.09070722553007526,
+      "grad_norm": 0.19088499248027802,
+      "learning_rate": 4.9755821953357144e-05,
+      "loss": 1.3774,
+      "step": 577
+    },
+    {
+      "epoch": 0.0908644304270078,
+      "grad_norm": 0.2181147336959839,
+      "learning_rate": 4.975495978471336e-05,
+      "loss": 1.3364,
+      "step": 578
+    },
+    {
+      "epoch": 0.09102163532394034,
+      "grad_norm": 0.18012750148773193,
+      "learning_rate": 4.975409610413082e-05,
+      "loss": 1.3852,
+      "step": 579
+    },
+    {
+      "epoch": 0.09117884022087289,
+      "grad_norm": 0.18108834326267242,
+      "learning_rate": 4.975323091166227e-05,
+      "loss": 1.3214,
+      "step": 580
+    },
+    {
+      "epoch": 0.09133604511780542,
+      "grad_norm": 0.25102898478507996,
+      "learning_rate": 4.975236420736056e-05,
+      "loss": 1.3199,
+      "step": 581
+    },
+    {
+      "epoch": 0.09149325001473796,
+      "grad_norm": 0.20121383666992188,
+      "learning_rate": 4.9751495991278626e-05,
+      "loss": 1.3328,
+      "step": 582
+    },
+    {
+      "epoch": 0.0916504549116705,
+      "grad_norm": 0.24183815717697144,
+      "learning_rate": 4.975062626346948e-05,
+      "loss": 1.3881,
+      "step": 583
+    },
+    {
+      "epoch": 0.09180765980860303,
+      "grad_norm": 0.23274902999401093,
+      "learning_rate": 4.974975502398626e-05,
+      "loss": 1.3674,
+      "step": 584
+    },
+    {
+      "epoch": 0.09196486470553558,
+      "grad_norm": 0.224375382065773,
+      "learning_rate": 4.9748882272882165e-05,
+      "loss": 1.362,
+      "step": 585
+    },
+    {
+      "epoch": 0.09212206960246812,
+      "grad_norm": 0.2743482291698456,
+      "learning_rate": 4.97480080102105e-05,
+      "loss": 1.3028,
+      "step": 586
+    },
+    {
+      "epoch": 0.09227927449940065,
+      "grad_norm": 0.30631452798843384,
+      "learning_rate": 4.974713223602467e-05,
+      "loss": 1.3541,
+      "step": 587
+    },
+    {
+      "epoch": 0.09243647939633319,
+      "grad_norm": 0.1999395489692688,
+      "learning_rate": 4.9746254950378166e-05,
+      "loss": 1.3515,
+      "step": 588
+    },
+    {
+      "epoch": 0.09259368429326574,
+      "grad_norm": 0.3005799353122711,
+      "learning_rate": 4.974537615332455e-05,
+      "loss": 1.3872,
+      "step": 589
+    },
+    {
+      "epoch": 0.09275088919019828,
+      "grad_norm": 0.21795117855072021,
+      "learning_rate": 4.9744495844917524e-05,
+      "loss": 1.2804,
+      "step": 590
+    },
+    {
+      "epoch": 0.09290809408713081,
+      "grad_norm": 0.2832283675670624,
+      "learning_rate": 4.9743614025210825e-05,
+      "loss": 1.3209,
+      "step": 591
+    },
+    {
+      "epoch": 0.09306529898406335,
+      "grad_norm": 0.21391350030899048,
+      "learning_rate": 4.9742730694258334e-05,
+      "loss": 1.3041,
+      "step": 592
+    },
+    {
+      "epoch": 0.0932225038809959,
+      "grad_norm": 0.21651242673397064,
+      "learning_rate": 4.974184585211399e-05,
+      "loss": 1.2529,
+      "step": 593
+    },
+    {
+      "epoch": 0.09337970877792844,
+      "grad_norm": 0.22796374559402466,
+      "learning_rate": 4.974095949883183e-05,
+      "loss": 1.3999,
+      "step": 594
+    },
+    {
+      "epoch": 0.09353691367486097,
+      "grad_norm": 0.21013247966766357,
+      "learning_rate": 4.9740071634466e-05,
+      "loss": 1.3626,
+      "step": 595
+    },
+    {
+      "epoch": 0.09369411857179351,
+      "grad_norm": 0.31589969992637634,
+      "learning_rate": 4.973918225907073e-05,
+      "loss": 1.4096,
+      "step": 596
+    },
+    {
+      "epoch": 0.09385132346872604,
+      "grad_norm": 0.2923184931278229,
+      "learning_rate": 4.973829137270033e-05,
+      "loss": 1.2116,
+      "step": 597
+    },
+    {
+      "epoch": 0.0940085283656586,
+      "grad_norm": 0.2147187739610672,
+      "learning_rate": 4.9737398975409224e-05,
+      "loss": 1.3909,
+      "step": 598
+    },
+    {
+      "epoch": 0.09416573326259113,
+      "grad_norm": 0.20287127792835236,
+      "learning_rate": 4.9736505067251896e-05,
+      "loss": 1.3621,
+      "step": 599
+    },
+    {
+      "epoch": 0.09432293815952367,
+      "grad_norm": 0.24703876674175262,
+      "learning_rate": 4.9735609648282965e-05,
+      "loss": 1.3525,
+      "step": 600
+    },
+    {
+      "epoch": 0.0944801430564562,
+      "grad_norm": 0.25060412287712097,
+      "learning_rate": 4.97347127185571e-05,
+      "loss": 1.36,
+      "step": 601
+    },
+    {
+      "epoch": 0.09463734795338875,
+      "grad_norm": 0.214557945728302,
+      "learning_rate": 4.9733814278129096e-05,
+      "loss": 1.4372,
+      "step": 602
+    },
+    {
+      "epoch": 0.09479455285032129,
+      "grad_norm": 0.1984785795211792,
+      "learning_rate": 4.9732914327053825e-05,
+      "loss": 1.3191,
+      "step": 603
+    },
+    {
+      "epoch": 0.09495175774725383,
+      "grad_norm": 0.2099440097808838,
+      "learning_rate": 4.9732012865386244e-05,
+      "loss": 1.313,
+      "step": 604
+    },
+    {
+      "epoch": 0.09510896264418636,
+      "grad_norm": 0.20393683016300201,
+      "learning_rate": 4.9731109893181423e-05,
+      "loss": 1.3465,
+      "step": 605
+    },
+    {
+      "epoch": 0.09526616754111891,
+      "grad_norm": 0.25346165895462036,
+      "learning_rate": 4.97302054104945e-05,
+      "loss": 1.3379,
+      "step": 606
+    },
+    {
+      "epoch": 0.09542337243805145,
+      "grad_norm": 0.21876423060894012,
+      "learning_rate": 4.9729299417380725e-05,
+      "loss": 1.2746,
+      "step": 607
+    },
+    {
+      "epoch": 0.09558057733498398,
+      "grad_norm": 0.21032990515232086,
+      "learning_rate": 4.9728391913895436e-05,
+      "loss": 1.3215,
+      "step": 608
+    },
+    {
+      "epoch": 0.09573778223191652,
+      "grad_norm": 0.2550762891769409,
+      "learning_rate": 4.9727482900094044e-05,
+      "loss": 1.3239,
+      "step": 609
+    },
+    {
+      "epoch": 0.09589498712884906,
+      "grad_norm": 0.31706327199935913,
+      "learning_rate": 4.972657237603208e-05,
+      "loss": 1.3467,
+      "step": 610
+    },
+    {
+      "epoch": 0.09605219202578161,
+      "grad_norm": 0.17176879942417145,
+      "learning_rate": 4.972566034176516e-05,
+      "loss": 1.3815,
+      "step": 611
+    },
+    {
+      "epoch": 0.09620939692271414,
+      "grad_norm": 0.22620820999145508,
+      "learning_rate": 4.972474679734898e-05,
+      "loss": 1.2593,
+      "step": 612
+    },
+    {
+      "epoch": 0.09636660181964668,
+      "grad_norm": 0.18735802173614502,
+      "learning_rate": 4.9723831742839334e-05,
+      "loss": 1.424,
+      "step": 613
+    },
+    {
+      "epoch": 0.09652380671657922,
+      "grad_norm": 0.2582910656929016,
+      "learning_rate": 4.972291517829211e-05,
+      "loss": 1.2741,
+      "step": 614
+    },
+    {
+      "epoch": 0.09668101161351177,
+      "grad_norm": 0.19907522201538086,
+      "learning_rate": 4.97219971037633e-05,
+      "loss": 1.2045,
+      "step": 615
+    },
+    {
+      "epoch": 0.0968382165104443,
+      "grad_norm": 0.20451949536800385,
+      "learning_rate": 4.972107751930896e-05,
+      "loss": 1.3026,
+      "step": 616
+    },
+    {
+      "epoch": 0.09699542140737684,
+      "grad_norm": 0.29682090878486633,
+      "learning_rate": 4.972015642498527e-05,
+      "loss": 1.3789,
+      "step": 617
+    },
+    {
+      "epoch": 0.09715262630430938,
+      "grad_norm": 0.27210530638694763,
+      "learning_rate": 4.9719233820848476e-05,
+      "loss": 1.3968,
+      "step": 618
+    },
+    {
+      "epoch": 0.09730983120124193,
+      "grad_norm": 0.24241842329502106,
+      "learning_rate": 4.971830970695493e-05,
+      "loss": 1.2763,
+      "step": 619
+    },
+    {
+      "epoch": 0.09746703609817446,
+      "grad_norm": 0.2535828649997711,
+      "learning_rate": 4.9717384083361075e-05,
+      "loss": 1.3463,
+      "step": 620
+    },
+    {
+      "epoch": 0.097624240995107,
+      "grad_norm": 0.22121217846870422,
+      "learning_rate": 4.971645695012344e-05,
+      "loss": 1.3384,
+      "step": 621
+    },
+    {
+      "epoch": 0.09778144589203953,
+      "grad_norm": 0.28840744495391846,
+      "learning_rate": 4.971552830729866e-05,
+      "loss": 1.2418,
+      "step": 622
+    },
+    {
+      "epoch": 0.09793865078897207,
+      "grad_norm": 0.1682664453983307,
+      "learning_rate": 4.971459815494345e-05,
+      "loss": 1.3658,
+      "step": 623
+    },
+    {
+      "epoch": 0.09809585568590462,
+      "grad_norm": 0.24955761432647705,
+      "learning_rate": 4.971366649311461e-05,
+      "loss": 1.2372,
+      "step": 624
+    },
+    {
+      "epoch": 0.09825306058283716,
+      "grad_norm": 0.2756117582321167,
+      "learning_rate": 4.971273332186906e-05,
+      "loss": 1.3212,
+      "step": 625
+    },
+    {
+      "epoch": 0.09841026547976969,
+      "grad_norm": 0.2370867133140564,
+      "learning_rate": 4.971179864126377e-05,
+      "loss": 1.2879,
+      "step": 626
+    },
+    {
+      "epoch": 0.09856747037670223,
+      "grad_norm": 0.20566895604133606,
+      "learning_rate": 4.9710862451355846e-05,
+      "loss": 1.4243,
+      "step": 627
+    },
+    {
+      "epoch": 0.09872467527363478,
+      "grad_norm": 0.1923399120569229,
+      "learning_rate": 4.970992475220246e-05,
+      "loss": 1.2639,
+      "step": 628
+    },
+    {
+      "epoch": 0.09888188017056732,
+      "grad_norm": 0.17972147464752197,
+      "learning_rate": 4.9708985543860896e-05,
+      "loss": 1.3366,
+      "step": 629
+    },
+    {
+      "epoch": 0.09903908506749985,
+      "grad_norm": 0.1936875432729721,
+      "learning_rate": 4.97080448263885e-05,
+      "loss": 1.3496,
+      "step": 630
+    },
+    {
+      "epoch": 0.09919628996443239,
+      "grad_norm": 0.24409984052181244,
+      "learning_rate": 4.9707102599842735e-05,
+      "loss": 1.3268,
+      "step": 631
+    },
+    {
+      "epoch": 0.09935349486136494,
+      "grad_norm": 0.21084928512573242,
+      "learning_rate": 4.970615886428115e-05,
+      "loss": 1.3421,
+      "step": 632
+    },
+    {
+      "epoch": 0.09951069975829747,
+      "grad_norm": 0.21201804280281067,
+      "learning_rate": 4.970521361976138e-05,
+      "loss": 1.3189,
+      "step": 633
+    },
+    {
+      "epoch": 0.09966790465523001,
+      "grad_norm": 0.2698107063770294,
+      "learning_rate": 4.9704266866341156e-05,
+      "loss": 1.2193,
+      "step": 634
+    },
+    {
+      "epoch": 0.09982510955216255,
+      "grad_norm": 0.27072674036026,
+      "learning_rate": 4.970331860407831e-05,
+      "loss": 1.2694,
+      "step": 635
+    },
+    {
+      "epoch": 0.09998231444909508,
+      "grad_norm": 0.26514896750450134,
+      "learning_rate": 4.9702368833030754e-05,
+      "loss": 1.2175,
+      "step": 636
+    },
+    {
+      "epoch": 0.10013951934602763,
+      "grad_norm": 0.21645940840244293,
+      "learning_rate": 4.970141755325649e-05,
+      "loss": 1.3099,
+      "step": 637
+    },
+    {
+      "epoch": 0.10029672424296017,
+      "grad_norm": 0.27035385370254517,
+      "learning_rate": 4.970046476481363e-05,
+      "loss": 1.2723,
+      "step": 638
+    },
+    {
+      "epoch": 0.1004539291398927,
+      "grad_norm": 0.20999298989772797,
+      "learning_rate": 4.969951046776036e-05,
+      "loss": 1.369,
+      "step": 639
+    },
+    {
+      "epoch": 0.10061113403682524,
+      "grad_norm": 0.18554192781448364,
+      "learning_rate": 4.969855466215497e-05,
+      "loss": 1.3483,
+      "step": 640
+    },
+    {
+      "epoch": 0.10061113403682524,
+      "eval_loss": 1.314468502998352,
+      "eval_runtime": 2275.7115,
+      "eval_samples_per_second": 4.068,
+      "eval_steps_per_second": 2.034,
+      "step": 640
+    },
+    {
+      "epoch": 0.10076833893375779,
+      "grad_norm": 0.19117292761802673,
+      "learning_rate": 4.969759734805582e-05,
+      "loss": 1.3538,
+      "step": 641
+    },
+    {
+      "epoch": 0.10092554383069033,
+      "grad_norm": 0.21971918642520905,
+      "learning_rate": 4.969663852552141e-05,
+      "loss": 1.2827,
+      "step": 642
+    },
+    {
+      "epoch": 0.10108274872762286,
+      "grad_norm": 0.2663845121860504,
+      "learning_rate": 4.969567819461027e-05,
+      "loss": 1.3332,
+      "step": 643
+    },
+    {
+      "epoch": 0.1012399536245554,
+      "grad_norm": 0.23752686381340027,
+      "learning_rate": 4.9694716355381076e-05,
+      "loss": 1.2675,
+      "step": 644
+    },
+    {
+      "epoch": 0.10139715852148795,
+      "grad_norm": 0.1558876782655716,
+      "learning_rate": 4.9693753007892565e-05,
+      "loss": 1.3356,
+      "step": 645
+    },
+    {
+      "epoch": 0.10155436341842049,
+      "grad_norm": 0.2064114212989807,
+      "learning_rate": 4.969278815220356e-05,
+      "loss": 1.3261,
+      "step": 646
+    },
+    {
+      "epoch": 0.10171156831535302,
+      "grad_norm": 0.2371819168329239,
+      "learning_rate": 4.969182178837302e-05,
+      "loss": 1.2706,
+      "step": 647
+    },
+    {
+      "epoch": 0.10186877321228556,
+      "grad_norm": 0.22757107019424438,
+      "learning_rate": 4.969085391645994e-05,
+      "loss": 1.4035,
+      "step": 648
+    },
+    {
+      "epoch": 0.1020259781092181,
+      "grad_norm": 0.16831007599830627,
+      "learning_rate": 4.968988453652345e-05,
+      "loss": 1.3006,
+      "step": 649
+    },
+    {
+      "epoch": 0.10218318300615065,
+      "grad_norm": 0.1719575822353363,
+      "learning_rate": 4.968891364862275e-05,
+      "loss": 1.2439,
+      "step": 650
+    },
+    {
+      "epoch": 0.10234038790308318,
+      "grad_norm": 0.27235090732574463,
+      "learning_rate": 4.9687941252817144e-05,
+      "loss": 1.3065,
+      "step": 651
+    },
+    {
+      "epoch": 0.10249759280001572,
+      "grad_norm": 0.25622984766960144,
+      "learning_rate": 4.968696734916601e-05,
+      "loss": 1.2908,
+      "step": 652
+    },
+    {
+      "epoch": 0.10265479769694826,
+      "grad_norm": 0.22526390850543976,
+      "learning_rate": 4.968599193772885e-05,
+      "loss": 1.3081,
+      "step": 653
+    },
+    {
+      "epoch": 0.1028120025938808,
+      "grad_norm": 0.2552133798599243,
+      "learning_rate": 4.968501501856522e-05,
+      "loss": 1.3292,
+      "step": 654
+    },
+    {
+      "epoch": 0.10296920749081334,
+      "grad_norm": 0.26533272862434387,
+      "learning_rate": 4.96840365917348e-05,
+      "loss": 1.3571,
+      "step": 655
+    },
+    {
+      "epoch": 0.10312641238774588,
+      "grad_norm": 0.29065170884132385,
+      "learning_rate": 4.968305665729732e-05,
+      "loss": 1.2799,
+      "step": 656
+    },
+    {
+      "epoch": 0.10328361728467841,
+      "grad_norm": 0.27552661299705505,
+      "learning_rate": 4.968207521531267e-05,
+      "loss": 1.2262,
+      "step": 657
+    },
+    {
+      "epoch": 0.10344082218161096,
+      "grad_norm": 1.929308533668518,
+      "learning_rate": 4.9681092265840775e-05,
+      "loss": 1.2027,
+      "step": 658
+    },
+    {
+      "epoch": 0.1035980270785435,
+      "grad_norm": 0.2610799968242645,
+      "learning_rate": 4.968010780894167e-05,
+      "loss": 1.3527,
+      "step": 659
+    },
+    {
+      "epoch": 0.10375523197547604,
+      "grad_norm": 0.28388604521751404,
+      "learning_rate": 4.967912184467547e-05,
+      "loss": 1.2989,
+      "step": 660
+    },
+    {
+      "epoch": 0.10391243687240857,
+      "grad_norm": 0.21056891977787018,
+      "learning_rate": 4.9678134373102415e-05,
+      "loss": 1.2748,
+      "step": 661
+    },
+    {
+      "epoch": 0.10406964176934111,
+      "grad_norm": 0.268331378698349,
+      "learning_rate": 4.967714539428281e-05,
+      "loss": 1.3712,
+      "step": 662
+    },
+    {
+      "epoch": 0.10422684666627366,
+      "grad_norm": 0.28430554270744324,
+      "learning_rate": 4.967615490827705e-05,
+      "loss": 1.3641,
+      "step": 663
+    },
+    {
+      "epoch": 0.1043840515632062,
+      "grad_norm": 0.254165917634964,
+      "learning_rate": 4.9675162915145636e-05,
+      "loss": 1.3042,
+      "step": 664
+    },
+    {
+      "epoch": 0.10454125646013873,
+      "grad_norm": 0.19123367965221405,
+      "learning_rate": 4.967416941494914e-05,
+      "loss": 1.3613,
+      "step": 665
+    },
+    {
+      "epoch": 0.10469846135707127,
+      "grad_norm": 0.20710323750972748,
+      "learning_rate": 4.967317440774828e-05,
+      "loss": 1.2815,
+      "step": 666
+    },
+    {
+      "epoch": 0.10485566625400382,
+      "grad_norm": 0.2143716812133789,
+      "learning_rate": 4.967217789360379e-05,
+      "loss": 1.3136,
+      "step": 667
+    },
+    {
+      "epoch": 0.10501287115093635,
+      "grad_norm": 0.2556392550468445,
+      "learning_rate": 4.967117987257654e-05,
+      "loss": 1.384,
+      "step": 668
+    },
+    {
+      "epoch": 0.10517007604786889,
+      "grad_norm": 0.28254854679107666,
+      "learning_rate": 4.9670180344727505e-05,
+      "loss": 1.3218,
+      "step": 669
+    },
+    {
+      "epoch": 0.10532728094480143,
+      "grad_norm": 0.24643027782440186,
+      "learning_rate": 4.9669179310117706e-05,
+      "loss": 1.278,
+      "step": 670
+    },
+    {
+      "epoch": 0.10548448584173396,
+      "grad_norm": 0.34323665499687195,
+      "learning_rate": 4.9668176768808304e-05,
+      "loss": 1.2511,
+      "step": 671
+    },
+    {
+      "epoch": 0.10564169073866651,
+      "grad_norm": 0.2499508410692215,
+      "learning_rate": 4.966717272086052e-05,
+      "loss": 1.338,
+      "step": 672
+    },
+    {
+      "epoch": 0.10579889563559905,
+      "grad_norm": 0.2145325094461441,
+      "learning_rate": 4.966616716633567e-05,
+      "loss": 1.3304,
+      "step": 673
+    },
+    {
+      "epoch": 0.10595610053253159,
+      "grad_norm": 0.19230923056602478,
+      "learning_rate": 4.9665160105295185e-05,
+      "loss": 1.3535,
+      "step": 674
+    },
+    {
+      "epoch": 0.10611330542946412,
+      "grad_norm": 0.20243465900421143,
+      "learning_rate": 4.966415153780056e-05,
+      "loss": 1.3118,
+      "step": 675
+    },
+    {
+      "epoch": 0.10627051032639667,
+      "grad_norm": 0.24927914142608643,
+      "learning_rate": 4.966314146391341e-05,
+      "loss": 1.3136,
+      "step": 676
+    },
+    {
+      "epoch": 0.10642771522332921,
+      "grad_norm": 0.21791934967041016,
+      "learning_rate": 4.9662129883695406e-05,
+      "loss": 1.3314,
+      "step": 677
+    },
+    {
+      "epoch": 0.10658492012026174,
+      "grad_norm": 0.24318841099739075,
+      "learning_rate": 4.966111679720835e-05,
+      "loss": 1.3929,
+      "step": 678
+    },
+    {
+      "epoch": 0.10674212501719428,
+      "grad_norm": 0.2829376757144928,
+      "learning_rate": 4.966010220451411e-05,
+      "loss": 1.3232,
+      "step": 679
+    },
+    {
+      "epoch": 0.10689932991412683,
+      "grad_norm": 0.2353716641664505,
+      "learning_rate": 4.965908610567465e-05,
+      "loss": 1.2851,
+      "step": 680
+    },
+    {
+      "epoch": 0.10705653481105937,
+      "grad_norm": 0.2615984380245209,
+      "learning_rate": 4.965806850075203e-05,
+      "loss": 1.2552,
+      "step": 681
+    },
+    {
+      "epoch": 0.1072137397079919,
+      "grad_norm": 0.23773109912872314,
+      "learning_rate": 4.965704938980841e-05,
+      "loss": 1.2961,
+      "step": 682
+    },
+    {
+      "epoch": 0.10737094460492444,
+      "grad_norm": 0.2622957229614258,
+      "learning_rate": 4.9656028772906014e-05,
+      "loss": 1.3073,
+      "step": 683
+    },
+    {
+      "epoch": 0.10752814950185698,
+      "grad_norm": 0.24974018335342407,
+      "learning_rate": 4.965500665010721e-05,
+      "loss": 1.2774,
+      "step": 684
+    },
+    {
+      "epoch": 0.10768535439878953,
+      "grad_norm": 0.17124338448047638,
+      "learning_rate": 4.9653983021474395e-05,
+      "loss": 1.4159,
+      "step": 685
+    },
+    {
+      "epoch": 0.10784255929572206,
+      "grad_norm": 0.16673363745212555,
+      "learning_rate": 4.96529578870701e-05,
+      "loss": 1.3748,
+      "step": 686
+    },
+    {
+      "epoch": 0.1079997641926546,
+      "grad_norm": 0.25368422269821167,
+      "learning_rate": 4.965193124695693e-05,
+      "loss": 1.3958,
+      "step": 687
+    },
+    {
+      "epoch": 0.10815696908958713,
+      "grad_norm": 0.22910015285015106,
+      "learning_rate": 4.96509031011976e-05,
+      "loss": 1.3259,
+      "step": 688
+    },
+    {
+      "epoch": 0.10831417398651969,
+      "grad_norm": 0.277851939201355,
+      "learning_rate": 4.96498734498549e-05,
+      "loss": 1.3254,
+      "step": 689
+    },
+    {
+      "epoch": 0.10847137888345222,
+      "grad_norm": 0.32443082332611084,
+      "learning_rate": 4.964884229299172e-05,
+      "loss": 1.3007,
+      "step": 690
+    },
+    {
+      "epoch": 0.10862858378038476,
+      "grad_norm": 0.20710885524749756,
+      "learning_rate": 4.964780963067102e-05,
+      "loss": 1.3297,
+      "step": 691
+    },
+    {
+      "epoch": 0.1087857886773173,
+      "grad_norm": 0.25522252917289734,
+      "learning_rate": 4.96467754629559e-05,
+      "loss": 1.2487,
+      "step": 692
+    },
+    {
+      "epoch": 0.10894299357424984,
+      "grad_norm": 0.3286147713661194,
+      "learning_rate": 4.9645739789909504e-05,
+      "loss": 1.2255,
+      "step": 693
+    },
+    {
+      "epoch": 0.10910019847118238,
+      "grad_norm": 0.3795601725578308,
+      "learning_rate": 4.964470261159509e-05,
+      "loss": 1.2725,
+      "step": 694
+    },
+    {
+      "epoch": 0.10925740336811492,
+      "grad_norm": 0.3112131655216217,
+      "learning_rate": 4.964366392807602e-05,
+      "loss": 1.252,
+      "step": 695
+    },
+    {
+      "epoch": 0.10941460826504745,
+      "grad_norm": 0.2891729176044464,
+      "learning_rate": 4.964262373941571e-05,
+      "loss": 1.3377,
+      "step": 696
+    },
+    {
+      "epoch": 0.10957181316197999,
+      "grad_norm": 0.26973745226860046,
+      "learning_rate": 4.96415820456777e-05,
+      "loss": 1.3186,
+      "step": 697
+    },
+    {
+      "epoch": 0.10972901805891254,
+      "grad_norm": 0.2832094430923462,
+      "learning_rate": 4.964053884692562e-05,
+      "loss": 1.3248,
+      "step": 698
+    },
+    {
+      "epoch": 0.10988622295584508,
+      "grad_norm": 0.2840999960899353,
+      "learning_rate": 4.963949414322318e-05,
+      "loss": 1.2677,
+      "step": 699
+    },
+    {
+      "epoch": 0.11004342785277761,
+      "grad_norm": 0.2891542911529541,
+      "learning_rate": 4.963844793463418e-05,
+      "loss": 1.3274,
+      "step": 700
+    },
+    {
+      "epoch": 0.11020063274971015,
+      "grad_norm": 0.23569005727767944,
+      "learning_rate": 4.963740022122252e-05,
+      "loss": 1.2259,
+      "step": 701
+    },
+    {
+      "epoch": 0.1103578376466427,
+      "grad_norm": 0.2174285501241684,
+      "learning_rate": 4.963635100305221e-05,
+      "loss": 1.2785,
+      "step": 702
+    },
+    {
+      "epoch": 0.11051504254357523,
+      "grad_norm": 0.2753438651561737,
+      "learning_rate": 4.96353002801873e-05,
+      "loss": 1.3263,
+      "step": 703
+    },
+    {
+      "epoch": 0.11067224744050777,
+      "grad_norm": 0.21094419062137604,
+      "learning_rate": 4.963424805269198e-05,
+      "loss": 1.2439,
+      "step": 704
+    },
+    {
+      "epoch": 0.1108294523374403,
+      "grad_norm": 0.20501388609409332,
+      "learning_rate": 4.963319432063052e-05,
+      "loss": 1.3091,
+      "step": 705
+    },
+    {
+      "epoch": 0.11098665723437286,
+      "grad_norm": 0.2041424810886383,
+      "learning_rate": 4.963213908406728e-05,
+      "loss": 1.2951,
+      "step": 706
+    },
+    {
+      "epoch": 0.1111438621313054,
+      "grad_norm": 0.24955442547798157,
+      "learning_rate": 4.963108234306669e-05,
+      "loss": 1.2208,
+      "step": 707
+    },
+    {
+      "epoch": 0.11130106702823793,
+      "grad_norm": 0.39431118965148926,
+      "learning_rate": 4.9630024097693314e-05,
+      "loss": 1.306,
+      "step": 708
+    },
+    {
+      "epoch": 0.11145827192517047,
+      "grad_norm": 0.24803434312343597,
+      "learning_rate": 4.962896434801178e-05,
+      "loss": 1.2951,
+      "step": 709
+    },
+    {
+      "epoch": 0.111615476822103,
+      "grad_norm": 0.2736116349697113,
+      "learning_rate": 4.962790309408681e-05,
+      "loss": 1.3245,
+      "step": 710
+    },
+    {
+      "epoch": 0.11177268171903555,
+      "grad_norm": 0.24502034485340118,
+      "learning_rate": 4.9626840335983215e-05,
+      "loss": 1.2961,
+      "step": 711
+    },
+    {
+      "epoch": 0.11192988661596809,
+      "grad_norm": 0.24158692359924316,
+      "learning_rate": 4.962577607376592e-05,
+      "loss": 1.2387,
+      "step": 712
+    },
+    {
+      "epoch": 0.11208709151290062,
+      "grad_norm": 0.24977251887321472,
+      "learning_rate": 4.962471030749991e-05,
+      "loss": 1.2976,
+      "step": 713
+    },
+    {
+      "epoch": 0.11224429640983316,
+      "grad_norm": 0.15401019155979156,
+      "learning_rate": 4.962364303725029e-05,
+      "loss": 1.2684,
+      "step": 714
+    },
+    {
+      "epoch": 0.11240150130676571,
+      "grad_norm": 0.2611544132232666,
+      "learning_rate": 4.962257426308224e-05,
+      "loss": 1.2928,
+      "step": 715
+    },
+    {
+      "epoch": 0.11255870620369825,
+      "grad_norm": 0.434600830078125,
+      "learning_rate": 4.962150398506103e-05,
+      "loss": 1.3657,
+      "step": 716
+    },
+    {
+      "epoch": 0.11271591110063078,
+      "grad_norm": 0.2896519601345062,
+      "learning_rate": 4.9620432203252045e-05,
+      "loss": 1.3055,
+      "step": 717
+    },
+    {
+      "epoch": 0.11287311599756332,
+      "grad_norm": 0.1891547590494156,
+      "learning_rate": 4.961935891772073e-05,
+      "loss": 1.3355,
+      "step": 718
+    },
+    {
+      "epoch": 0.11303032089449587,
+      "grad_norm": 0.2223133146762848,
+      "learning_rate": 4.9618284128532644e-05,
+      "loss": 1.2939,
+      "step": 719
+    },
+    {
+      "epoch": 0.1131875257914284,
+      "grad_norm": 0.27313077449798584,
+      "learning_rate": 4.961720783575343e-05,
+      "loss": 1.2596,
+      "step": 720
+    },
+    {
+      "epoch": 0.11334473068836094,
+      "grad_norm": 0.24807053804397583,
+      "learning_rate": 4.961613003944883e-05,
+      "loss": 1.2851,
+      "step": 721
+    },
+    {
+      "epoch": 0.11350193558529348,
+      "grad_norm": 0.2343195378780365,
+      "learning_rate": 4.9615050739684656e-05,
+      "loss": 1.2899,
+      "step": 722
+    },
+    {
+      "epoch": 0.11365914048222601,
+      "grad_norm": 0.229730024933815,
+      "learning_rate": 4.961396993652684e-05,
+      "loss": 1.3118,
+      "step": 723
+    },
+    {
+      "epoch": 0.11381634537915856,
+      "grad_norm": 0.2397170215845108,
+      "learning_rate": 4.9612887630041394e-05,
+      "loss": 1.2148,
+      "step": 724
+    },
+    {
+      "epoch": 0.1139735502760911,
+      "grad_norm": 0.2167958915233612,
+      "learning_rate": 4.9611803820294414e-05,
+      "loss": 1.2597,
+      "step": 725
+    },
+    {
+      "epoch": 0.11413075517302364,
+      "grad_norm": 0.21318721771240234,
+      "learning_rate": 4.961071850735209e-05,
+      "loss": 1.3949,
+      "step": 726
+    },
+    {
+      "epoch": 0.11428796006995617,
+      "grad_norm": 0.21988382935523987,
+      "learning_rate": 4.960963169128073e-05,
+      "loss": 1.3196,
+      "step": 727
+    },
+    {
+      "epoch": 0.11444516496688872,
+      "grad_norm": 0.17555692791938782,
+      "learning_rate": 4.96085433721467e-05,
+      "loss": 1.3661,
+      "step": 728
+    },
+    {
+      "epoch": 0.11460236986382126,
+      "grad_norm": 0.3545222282409668,
+      "learning_rate": 4.960745355001647e-05,
+      "loss": 1.2659,
+      "step": 729
+    },
+    {
+      "epoch": 0.1147595747607538,
+      "grad_norm": 0.3196569085121155,
+      "learning_rate": 4.960636222495659e-05,
+      "loss": 1.2893,
+      "step": 730
+    },
+    {
+      "epoch": 0.11491677965768633,
+      "grad_norm": 0.2241334766149521,
+      "learning_rate": 4.960526939703374e-05,
+      "loss": 1.2155,
+      "step": 731
+    },
+    {
+      "epoch": 0.11507398455461888,
+      "grad_norm": 0.26543980836868286,
+      "learning_rate": 4.960417506631465e-05,
+      "loss": 1.3615,
+      "step": 732
+    },
+    {
+      "epoch": 0.11523118945155142,
+      "grad_norm": 0.21146585047245026,
+      "learning_rate": 4.960307923286616e-05,
+      "loss": 1.3516,
+      "step": 733
+    },
+    {
+      "epoch": 0.11538839434848396,
+      "grad_norm": 0.18095079064369202,
+      "learning_rate": 4.960198189675519e-05,
+      "loss": 1.3581,
+      "step": 734
+    },
+    {
+      "epoch": 0.11554559924541649,
+      "grad_norm": 0.26687100529670715,
+      "learning_rate": 4.9600883058048775e-05,
+      "loss": 1.1971,
+      "step": 735
+    },
+    {
+      "epoch": 0.11570280414234903,
+      "grad_norm": 0.2271047830581665,
+      "learning_rate": 4.959978271681402e-05,
+      "loss": 1.1867,
+      "step": 736
+    },
+    {
+      "epoch": 0.11586000903928158,
+      "grad_norm": 0.2102867215871811,
+      "learning_rate": 4.959868087311814e-05,
+      "loss": 1.2749,
+      "step": 737
+    },
+    {
+      "epoch": 0.11601721393621411,
+      "grad_norm": 0.2752761244773865,
+      "learning_rate": 4.9597577527028424e-05,
+      "loss": 1.1753,
+      "step": 738
+    },
+    {
+      "epoch": 0.11617441883314665,
+      "grad_norm": 0.22385725378990173,
+      "learning_rate": 4.959647267861226e-05,
+      "loss": 1.343,
+      "step": 739
+    },
+    {
+      "epoch": 0.11633162373007919,
+      "grad_norm": 0.2597412168979645,
+      "learning_rate": 4.959536632793712e-05,
+      "loss": 1.2539,
+      "step": 740
+    },
+    {
+      "epoch": 0.11648882862701174,
+      "grad_norm": 0.27975237369537354,
+      "learning_rate": 4.959425847507059e-05,
+      "loss": 1.2883,
+      "step": 741
+    },
+    {
+      "epoch": 0.11664603352394427,
+      "grad_norm": 0.29127049446105957,
+      "learning_rate": 4.959314912008033e-05,
+      "loss": 1.3139,
+      "step": 742
+    },
+    {
+      "epoch": 0.11680323842087681,
+      "grad_norm": 0.19929318130016327,
+      "learning_rate": 4.9592038263034094e-05,
+      "loss": 1.271,
+      "step": 743
+    },
+    {
+      "epoch": 0.11696044331780935,
+      "grad_norm": 0.23164550960063934,
+      "learning_rate": 4.9590925903999716e-05,
+      "loss": 1.3359,
+      "step": 744
+    },
+    {
+      "epoch": 0.1171176482147419,
+      "grad_norm": 0.27876612544059753,
+      "learning_rate": 4.958981204304516e-05,
+      "loss": 1.2568,
+      "step": 745
+    },
+    {
+      "epoch": 0.11727485311167443,
+      "grad_norm": 0.2459796965122223,
+      "learning_rate": 4.9588696680238435e-05,
+      "loss": 1.2426,
+      "step": 746
+    },
+    {
+      "epoch": 0.11743205800860697,
+      "grad_norm": 0.2039456069469452,
+      "learning_rate": 4.958757981564767e-05,
+      "loss": 1.2681,
+      "step": 747
+    },
+    {
+      "epoch": 0.1175892629055395,
+      "grad_norm": 0.24796408414840698,
+      "learning_rate": 4.958646144934108e-05,
+      "loss": 1.257,
+      "step": 748
+    },
+    {
+      "epoch": 0.11774646780247204,
+      "grad_norm": 0.2779620289802551,
+      "learning_rate": 4.958534158138697e-05,
+      "loss": 1.2933,
+      "step": 749
+    },
+    {
+      "epoch": 0.11790367269940459,
+      "grad_norm": 0.20878851413726807,
+      "learning_rate": 4.9584220211853735e-05,
+      "loss": 1.2902,
+      "step": 750
+    },
+    {
+      "epoch": 0.11806087759633713,
+      "grad_norm": 0.24720412492752075,
+      "learning_rate": 4.958309734080987e-05,
+      "loss": 1.203,
+      "step": 751
+    },
+    {
+      "epoch": 0.11821808249326966,
+      "grad_norm": 0.287654846906662,
+      "learning_rate": 4.9581972968323956e-05,
+      "loss": 1.3141,
+      "step": 752
+    },
+    {
+      "epoch": 0.1183752873902022,
+      "grad_norm": 0.23071719706058502,
+      "learning_rate": 4.958084709446466e-05,
+      "loss": 1.3145,
+      "step": 753
+    },
+    {
+      "epoch": 0.11853249228713475,
+      "grad_norm": 0.21027110517024994,
+      "learning_rate": 4.9579719719300746e-05,
+      "loss": 1.2893,
+      "step": 754
+    },
+    {
+      "epoch": 0.11868969718406729,
+      "grad_norm": 0.17173202335834503,
+      "learning_rate": 4.9578590842901066e-05,
+      "loss": 1.2618,
+      "step": 755
+    },
+    {
+      "epoch": 0.11884690208099982,
+      "grad_norm": 0.24606984853744507,
+      "learning_rate": 4.957746046533457e-05,
+      "loss": 1.1904,
+      "step": 756
+    },
+    {
+      "epoch": 0.11900410697793236,
+      "grad_norm": 0.248653382062912,
+      "learning_rate": 4.957632858667031e-05,
+      "loss": 1.331,
+      "step": 757
+    },
+    {
+      "epoch": 0.11916131187486491,
+      "grad_norm": 0.1904144436120987,
+      "learning_rate": 4.9575195206977406e-05,
+      "loss": 1.3303,
+      "step": 758
+    },
+    {
+      "epoch": 0.11931851677179744,
+      "grad_norm": 0.39540621638298035,
+      "learning_rate": 4.9574060326325075e-05,
+      "loss": 1.3455,
+      "step": 759
+    },
+    {
+      "epoch": 0.11947572166872998,
+      "grad_norm": 0.20992301404476166,
+      "learning_rate": 4.957292394478265e-05,
+      "loss": 1.2911,
+      "step": 760
+    },
+    {
+      "epoch": 0.11963292656566252,
+      "grad_norm": 0.23418502509593964,
+      "learning_rate": 4.957178606241951e-05,
+      "loss": 1.35,
+      "step": 761
+    },
+    {
+      "epoch": 0.11979013146259505,
+      "grad_norm": 0.24480225145816803,
+      "learning_rate": 4.957064667930517e-05,
+      "loss": 1.2138,
+      "step": 762
+    },
+    {
+      "epoch": 0.1199473363595276,
+      "grad_norm": 0.22909322381019592,
+      "learning_rate": 4.956950579550922e-05,
+      "loss": 1.1915,
+      "step": 763
+    },
+    {
+      "epoch": 0.12010454125646014,
+      "grad_norm": 0.16839763522148132,
+      "learning_rate": 4.956836341110134e-05,
+      "loss": 1.234,
+      "step": 764
+    },
+    {
+      "epoch": 0.12026174615339268,
+      "grad_norm": 0.2291131466627121,
+      "learning_rate": 4.956721952615129e-05,
+      "loss": 1.2964,
+      "step": 765
+    },
+    {
+      "epoch": 0.12041895105032521,
+      "grad_norm": 0.2606765329837799,
+      "learning_rate": 4.956607414072895e-05,
+      "loss": 1.2785,
+      "step": 766
+    },
+    {
+      "epoch": 0.12057615594725776,
+      "grad_norm": 0.24100011587142944,
+      "learning_rate": 4.956492725490426e-05,
+      "loss": 1.2389,
+      "step": 767
+    },
+    {
+      "epoch": 0.1207333608441903,
+      "grad_norm": 0.2868693172931671,
+      "learning_rate": 4.956377886874729e-05,
+      "loss": 1.3852,
+      "step": 768
+    },
+    {
+      "epoch": 0.12089056574112284,
+      "grad_norm": 0.29049259424209595,
+      "learning_rate": 4.956262898232816e-05,
+      "loss": 1.1511,
+      "step": 769
+    },
+    {
+      "epoch": 0.12104777063805537,
+      "grad_norm": 0.31396448612213135,
+      "learning_rate": 4.9561477595717106e-05,
+      "loss": 1.2687,
+      "step": 770
+    },
+    {
+      "epoch": 0.12120497553498792,
+      "grad_norm": 0.3348733186721802,
+      "learning_rate": 4.956032470898445e-05,
+      "loss": 1.1933,
+      "step": 771
+    },
+    {
+      "epoch": 0.12136218043192046,
+      "grad_norm": 0.2009342461824417,
+      "learning_rate": 4.955917032220061e-05,
+      "loss": 1.3299,
+      "step": 772
+    },
+    {
+      "epoch": 0.121519385328853,
+      "grad_norm": 0.2037377655506134,
+      "learning_rate": 4.9558014435436084e-05,
+      "loss": 1.3208,
+      "step": 773
+    },
+    {
+      "epoch": 0.12167659022578553,
+      "grad_norm": 0.3118877410888672,
+      "learning_rate": 4.955685704876147e-05,
+      "loss": 1.1927,
+      "step": 774
+    },
+    {
+      "epoch": 0.12183379512271807,
+      "grad_norm": 0.21884632110595703,
+      "learning_rate": 4.955569816224747e-05,
+      "loss": 1.2661,
+      "step": 775
+    },
+    {
+      "epoch": 0.12199100001965062,
+      "grad_norm": 0.25817862153053284,
+      "learning_rate": 4.9554537775964846e-05,
+      "loss": 1.3077,
+      "step": 776
+    },
+    {
+      "epoch": 0.12214820491658315,
+      "grad_norm": 0.27827751636505127,
+      "learning_rate": 4.955337588998449e-05,
+      "loss": 1.2709,
+      "step": 777
+    },
+    {
+      "epoch": 0.12230540981351569,
+      "grad_norm": 0.30520737171173096,
+      "learning_rate": 4.955221250437735e-05,
+      "loss": 1.2407,
+      "step": 778
+    },
+    {
+      "epoch": 0.12246261471044823,
+      "grad_norm": 0.21729423105716705,
+      "learning_rate": 4.9551047619214473e-05,
+      "loss": 1.3392,
+      "step": 779
+    },
+    {
+      "epoch": 0.12261981960738078,
+      "grad_norm": 0.2408866286277771,
+      "learning_rate": 4.954988123456703e-05,
+      "loss": 1.215,
+      "step": 780
+    },
+    {
+      "epoch": 0.12277702450431331,
+      "grad_norm": 0.23833869397640228,
+      "learning_rate": 4.954871335050625e-05,
+      "loss": 1.3607,
+      "step": 781
+    },
+    {
+      "epoch": 0.12293422940124585,
+      "grad_norm": 0.27017349004745483,
+      "learning_rate": 4.954754396710345e-05,
+      "loss": 1.2662,
+      "step": 782
+    },
+    {
+      "epoch": 0.12309143429817838,
+      "grad_norm": 0.21869684755802155,
+      "learning_rate": 4.954637308443007e-05,
+      "loss": 1.2384,
+      "step": 783
+    },
+    {
+      "epoch": 0.12324863919511093,
+      "grad_norm": 0.18912911415100098,
+      "learning_rate": 4.9545200702557615e-05,
+      "loss": 1.2958,
+      "step": 784
+    },
+    {
+      "epoch": 0.12340584409204347,
+      "grad_norm": 0.27320876717567444,
+      "learning_rate": 4.954402682155768e-05,
+      "loss": 1.2546,
+      "step": 785
+    },
+    {
+      "epoch": 0.12356304898897601,
+      "grad_norm": 0.2938046455383301,
+      "learning_rate": 4.954285144150198e-05,
+      "loss": 1.3451,
+      "step": 786
+    },
+    {
+      "epoch": 0.12372025388590854,
+      "grad_norm": 0.18271508812904358,
+      "learning_rate": 4.954167456246229e-05,
+      "loss": 1.2239,
+      "step": 787
+    },
+    {
+      "epoch": 0.12387745878284108,
+      "grad_norm": 0.21799346804618835,
+      "learning_rate": 4.9540496184510495e-05,
+      "loss": 1.2471,
+      "step": 788
+    },
+    {
+      "epoch": 0.12403466367977363,
+      "grad_norm": 0.21574997901916504,
+      "learning_rate": 4.9539316307718564e-05,
+      "loss": 1.3137,
+      "step": 789
+    },
+    {
+      "epoch": 0.12419186857670617,
+      "grad_norm": 0.21586358547210693,
+      "learning_rate": 4.953813493215855e-05,
+      "loss": 1.2763,
+      "step": 790
+    },
+    {
+      "epoch": 0.1243490734736387,
+      "grad_norm": 0.2723408043384552,
+      "learning_rate": 4.953695205790262e-05,
+      "loss": 1.4148,
+      "step": 791
+    },
+    {
+      "epoch": 0.12450627837057124,
+      "grad_norm": 0.29501527547836304,
+      "learning_rate": 4.9535767685023026e-05,
+      "loss": 1.3093,
+      "step": 792
+    },
+    {
+      "epoch": 0.12466348326750379,
+      "grad_norm": 0.2884112000465393,
+      "learning_rate": 4.9534581813592086e-05,
+      "loss": 1.3276,
+      "step": 793
+    },
+    {
+      "epoch": 0.12482068816443632,
+      "grad_norm": 0.24246759712696075,
+      "learning_rate": 4.9533394443682234e-05,
+      "loss": 1.3203,
+      "step": 794
+    },
+    {
+      "epoch": 0.12497789306136886,
+      "grad_norm": 0.23493270576000214,
+      "learning_rate": 4.9532205575365995e-05,
+      "loss": 1.2567,
+      "step": 795
+    },
+    {
+      "epoch": 0.1251350979583014,
+      "grad_norm": 0.26456305384635925,
+      "learning_rate": 4.953101520871598e-05,
+      "loss": 1.3194,
+      "step": 796
+    },
+    {
+      "epoch": 0.12529230285523393,
+      "grad_norm": 0.18891221284866333,
+      "learning_rate": 4.952982334380489e-05,
+      "loss": 1.3041,
+      "step": 797
+    },
+    {
+      "epoch": 0.12544950775216648,
+      "grad_norm": 0.21460258960723877,
+      "learning_rate": 4.952862998070552e-05,
+      "loss": 1.2274,
+      "step": 798
+    },
+    {
+      "epoch": 0.125606712649099,
+      "grad_norm": 0.2832646667957306,
+      "learning_rate": 4.9527435119490753e-05,
+      "loss": 1.2009,
+      "step": 799
+    },
+    {
+      "epoch": 0.12576391754603156,
+      "grad_norm": 0.22183702886104584,
+      "learning_rate": 4.9526238760233576e-05,
+      "loss": 1.31,
+      "step": 800
+    },
+    {
+      "epoch": 0.12576391754603156,
+      "eval_loss": 1.2745521068572998,
+      "eval_runtime": 2292.1003,
+      "eval_samples_per_second": 4.039,
+      "eval_steps_per_second": 2.02,
+      "step": 800
+    },
+    {
+      "epoch": 0.1259211224429641,
+      "grad_norm": 0.21697020530700684,
+      "learning_rate": 4.9525040903007046e-05,
+      "loss": 1.3197,
+      "step": 801
+    },
+    {
+      "epoch": 0.12607832733989663,
+      "grad_norm": 0.36354196071624756,
+      "learning_rate": 4.952384154788433e-05,
+      "loss": 1.1926,
+      "step": 802
+    },
+    {
+      "epoch": 0.12623553223682918,
+      "grad_norm": 0.27054232358932495,
+      "learning_rate": 4.952264069493868e-05,
+      "loss": 1.3199,
+      "step": 803
+    },
+    {
+      "epoch": 0.12639273713376173,
+      "grad_norm": 0.2425469011068344,
+      "learning_rate": 4.952143834424344e-05,
+      "loss": 1.2906,
+      "step": 804
+    },
+    {
+      "epoch": 0.12654994203069425,
+      "grad_norm": 0.1988941729068756,
+      "learning_rate": 4.952023449587205e-05,
+      "loss": 1.3183,
+      "step": 805
+    },
+    {
+      "epoch": 0.1267071469276268,
+      "grad_norm": 0.2429157942533493,
+      "learning_rate": 4.951902914989802e-05,
+      "loss": 1.2497,
+      "step": 806
+    },
+    {
+      "epoch": 0.12686435182455932,
+      "grad_norm": 0.2704293727874756,
+      "learning_rate": 4.951782230639499e-05,
+      "loss": 1.3113,
+      "step": 807
+    },
+    {
+      "epoch": 0.12702155672149187,
+      "grad_norm": 0.31801360845565796,
+      "learning_rate": 4.951661396543664e-05,
+      "loss": 1.2354,
+      "step": 808
+    },
+    {
+      "epoch": 0.12717876161842442,
+      "grad_norm": 0.21358463168144226,
+      "learning_rate": 4.951540412709681e-05,
+      "loss": 1.3512,
+      "step": 809
+    },
+    {
+      "epoch": 0.12733596651535695,
+      "grad_norm": 0.24300484359264374,
+      "learning_rate": 4.951419279144936e-05,
+      "loss": 1.213,
+      "step": 810
+    },
+    {
+      "epoch": 0.1274931714122895,
+      "grad_norm": 0.39550015330314636,
+      "learning_rate": 4.951297995856828e-05,
+      "loss": 1.1872,
+      "step": 811
+    },
+    {
+      "epoch": 0.12765037630922202,
+      "grad_norm": 0.20150414109230042,
+      "learning_rate": 4.951176562852765e-05,
+      "loss": 1.3469,
+      "step": 812
+    },
+    {
+      "epoch": 0.12780758120615457,
+      "grad_norm": 0.2050725519657135,
+      "learning_rate": 4.951054980140164e-05,
+      "loss": 1.259,
+      "step": 813
+    },
+    {
+      "epoch": 0.12796478610308712,
+      "grad_norm": 0.23815183341503143,
+      "learning_rate": 4.950933247726451e-05,
+      "loss": 1.2961,
+      "step": 814
+    },
+    {
+      "epoch": 0.12812199100001964,
+      "grad_norm": 0.28224676847457886,
+      "learning_rate": 4.95081136561906e-05,
+      "loss": 1.2631,
+      "step": 815
+    },
+    {
+      "epoch": 0.1282791958969522,
+      "grad_norm": 0.294791042804718,
+      "learning_rate": 4.9506893338254353e-05,
+      "loss": 1.1834,
+      "step": 816
+    },
+    {
+      "epoch": 0.12843640079388474,
+      "grad_norm": 0.29148972034454346,
+      "learning_rate": 4.9505671523530306e-05,
+      "loss": 1.2573,
+      "step": 817
+    },
+    {
+      "epoch": 0.12859360569081726,
+      "grad_norm": 0.29371243715286255,
+      "learning_rate": 4.950444821209308e-05,
+      "loss": 1.4532,
+      "step": 818
+    },
+    {
+      "epoch": 0.12875081058774981,
+      "grad_norm": 0.2303713709115982,
+      "learning_rate": 4.9503223404017396e-05,
+      "loss": 1.2828,
+      "step": 819
+    },
+    {
+      "epoch": 0.12890801548468234,
+      "grad_norm": 0.24906295537948608,
+      "learning_rate": 4.9501997099378046e-05,
+      "loss": 1.2759,
+      "step": 820
+    },
+    {
+      "epoch": 0.1290652203816149,
+      "grad_norm": 0.1983998864889145,
+      "learning_rate": 4.950076929824994e-05,
+      "loss": 1.3111,
+      "step": 821
+    },
+    {
+      "epoch": 0.12922242527854744,
+      "grad_norm": 0.2079075276851654,
+      "learning_rate": 4.9499540000708064e-05,
+      "loss": 1.3416,
+      "step": 822
+    },
+    {
+      "epoch": 0.12937963017547996,
+      "grad_norm": 0.22548237442970276,
+      "learning_rate": 4.94983092068275e-05,
+      "loss": 1.3879,
+      "step": 823
+    },
+    {
+      "epoch": 0.1295368350724125,
+      "grad_norm": 0.2052278220653534,
+      "learning_rate": 4.949707691668343e-05,
+      "loss": 1.3347,
+      "step": 824
+    },
+    {
+      "epoch": 0.12969403996934503,
+      "grad_norm": 0.21978795528411865,
+      "learning_rate": 4.949584313035109e-05,
+      "loss": 1.1345,
+      "step": 825
+    },
+    {
+      "epoch": 0.12985124486627758,
+      "grad_norm": 0.18930193781852722,
+      "learning_rate": 4.9494607847905863e-05,
+      "loss": 1.319,
+      "step": 826
+    },
+    {
+      "epoch": 0.13000844976321013,
+      "grad_norm": 0.24538543820381165,
+      "learning_rate": 4.9493371069423176e-05,
+      "loss": 1.3103,
+      "step": 827
+    },
+    {
+      "epoch": 0.13016565466014265,
+      "grad_norm": 0.2874930799007416,
+      "learning_rate": 4.9492132794978586e-05,
+      "loss": 1.3388,
+      "step": 828
+    },
+    {
+      "epoch": 0.1303228595570752,
+      "grad_norm": 0.23338377475738525,
+      "learning_rate": 4.949089302464771e-05,
+      "loss": 1.2793,
+      "step": 829
+    },
+    {
+      "epoch": 0.13048006445400775,
+      "grad_norm": 0.23670902848243713,
+      "learning_rate": 4.948965175850626e-05,
+      "loss": 1.2808,
+      "step": 830
+    },
+    {
+      "epoch": 0.13063726935094028,
+      "grad_norm": 0.2617732584476471,
+      "learning_rate": 4.9488408996630066e-05,
+      "loss": 1.2641,
+      "step": 831
+    },
+    {
+      "epoch": 0.13079447424787283,
+      "grad_norm": 0.24584044516086578,
+      "learning_rate": 4.948716473909502e-05,
+      "loss": 1.2462,
+      "step": 832
+    },
+    {
+      "epoch": 0.13095167914480535,
+      "grad_norm": 0.2507297098636627,
+      "learning_rate": 4.948591898597712e-05,
+      "loss": 1.2211,
+      "step": 833
+    },
+    {
+      "epoch": 0.1311088840417379,
+      "grad_norm": 0.25439611077308655,
+      "learning_rate": 4.948467173735245e-05,
+      "loss": 1.2762,
+      "step": 834
+    },
+    {
+      "epoch": 0.13126608893867045,
+      "grad_norm": 0.19934779405593872,
+      "learning_rate": 4.948342299329719e-05,
+      "loss": 1.1798,
+      "step": 835
+    },
+    {
+      "epoch": 0.13142329383560297,
+      "grad_norm": 0.24154123663902283,
+      "learning_rate": 4.948217275388761e-05,
+      "loss": 1.2608,
+      "step": 836
+    },
+    {
+      "epoch": 0.13158049873253552,
+      "grad_norm": 0.2484877109527588,
+      "learning_rate": 4.948092101920006e-05,
+      "loss": 1.2466,
+      "step": 837
+    },
+    {
+      "epoch": 0.13173770362946804,
+      "grad_norm": 0.28683343529701233,
+      "learning_rate": 4.9479667789311e-05,
+      "loss": 1.1915,
+      "step": 838
+    },
+    {
+      "epoch": 0.1318949085264006,
+      "grad_norm": 0.21289369463920593,
+      "learning_rate": 4.9478413064296976e-05,
+      "loss": 1.2642,
+      "step": 839
+    },
+    {
+      "epoch": 0.13205211342333315,
+      "grad_norm": 0.22933778166770935,
+      "learning_rate": 4.947715684423461e-05,
+      "loss": 1.2182,
+      "step": 840
+    },
+    {
+      "epoch": 0.13220931832026567,
+      "grad_norm": 0.2507724463939667,
+      "learning_rate": 4.9475899129200635e-05,
+      "loss": 1.3089,
+      "step": 841
+    },
+    {
+      "epoch": 0.13236652321719822,
+      "grad_norm": 0.251770943403244,
+      "learning_rate": 4.947463991927187e-05,
+      "loss": 1.3194,
+      "step": 842
+    },
+    {
+      "epoch": 0.13252372811413077,
+      "grad_norm": 0.2533280849456787,
+      "learning_rate": 4.947337921452521e-05,
+      "loss": 1.2141,
+      "step": 843
+    },
+    {
+      "epoch": 0.1326809330110633,
+      "grad_norm": 0.26309993863105774,
+      "learning_rate": 4.9472117015037664e-05,
+      "loss": 1.2265,
+      "step": 844
+    },
+    {
+      "epoch": 0.13283813790799584,
+      "grad_norm": 0.29711806774139404,
+      "learning_rate": 4.9470853320886335e-05,
+      "loss": 1.2538,
+      "step": 845
+    },
+    {
+      "epoch": 0.13299534280492836,
+      "grad_norm": 0.24551883339881897,
+      "learning_rate": 4.9469588132148373e-05,
+      "loss": 1.2927,
+      "step": 846
+    },
+    {
+      "epoch": 0.1331525477018609,
+      "grad_norm": 0.28027257323265076,
+      "learning_rate": 4.946832144890108e-05,
+      "loss": 1.2712,
+      "step": 847
+    },
+    {
+      "epoch": 0.13330975259879346,
+      "grad_norm": 0.22099149227142334,
+      "learning_rate": 4.9467053271221804e-05,
+      "loss": 1.2095,
+      "step": 848
+    },
+    {
+      "epoch": 0.13346695749572599,
+      "grad_norm": 0.19661381840705872,
+      "learning_rate": 4.946578359918801e-05,
+      "loss": 1.2855,
+      "step": 849
+    },
+    {
+      "epoch": 0.13362416239265854,
+      "grad_norm": 0.22767631709575653,
+      "learning_rate": 4.946451243287723e-05,
+      "loss": 1.2932,
+      "step": 850
+    },
+    {
+      "epoch": 0.13378136728959106,
+      "grad_norm": 0.28008589148521423,
+      "learning_rate": 4.946323977236712e-05,
+      "loss": 1.2335,
+      "step": 851
+    },
+    {
+      "epoch": 0.1339385721865236,
+      "grad_norm": 0.2091825157403946,
+      "learning_rate": 4.94619656177354e-05,
+      "loss": 1.3151,
+      "step": 852
+    },
+    {
+      "epoch": 0.13409577708345616,
+      "grad_norm": 0.1978277713060379,
+      "learning_rate": 4.946068996905989e-05,
+      "loss": 1.3359,
+      "step": 853
+    },
+    {
+      "epoch": 0.13425298198038868,
+      "grad_norm": 0.21397674083709717,
+      "learning_rate": 4.9459412826418505e-05,
+      "loss": 1.2998,
+      "step": 854
+    },
+    {
+      "epoch": 0.13441018687732123,
+      "grad_norm": 0.30490776896476746,
+      "learning_rate": 4.945813418988925e-05,
+      "loss": 1.2607,
+      "step": 855
+    },
+    {
+      "epoch": 0.13456739177425378,
+      "grad_norm": 0.2896914780139923,
+      "learning_rate": 4.945685405955021e-05,
+      "loss": 1.2329,
+      "step": 856
+    },
+    {
+      "epoch": 0.1347245966711863,
+      "grad_norm": 0.1988048106431961,
+      "learning_rate": 4.945557243547958e-05,
+      "loss": 1.2877,
+      "step": 857
+    },
+    {
+      "epoch": 0.13488180156811885,
+      "grad_norm": 0.17888212203979492,
+      "learning_rate": 4.945428931775563e-05,
+      "loss": 1.2543,
+      "step": 858
+    },
+    {
+      "epoch": 0.13503900646505138,
+      "grad_norm": 0.2748056650161743,
+      "learning_rate": 4.945300470645673e-05,
+      "loss": 1.3461,
+      "step": 859
+    },
+    {
+      "epoch": 0.13519621136198393,
+      "grad_norm": 0.23218591511249542,
+      "learning_rate": 4.945171860166135e-05,
+      "loss": 1.2878,
+      "step": 860
+    },
+    {
+      "epoch": 0.13535341625891648,
+      "grad_norm": 0.33142325282096863,
+      "learning_rate": 4.9450431003448015e-05,
+      "loss": 1.294,
+      "step": 861
+    },
+    {
+      "epoch": 0.135510621155849,
+      "grad_norm": 0.2330816686153412,
+      "learning_rate": 4.944914191189539e-05,
+      "loss": 1.3593,
+      "step": 862
+    },
+    {
+      "epoch": 0.13566782605278155,
+      "grad_norm": 0.23989921808242798,
+      "learning_rate": 4.9447851327082204e-05,
+      "loss": 1.2879,
+      "step": 863
+    },
+    {
+      "epoch": 0.13582503094971407,
+      "grad_norm": 0.21358944475650787,
+      "learning_rate": 4.944655924908727e-05,
+      "loss": 1.222,
+      "step": 864
+    },
+    {
+      "epoch": 0.13598223584664662,
+      "grad_norm": 0.30434924364089966,
+      "learning_rate": 4.9445265677989515e-05,
+      "loss": 1.273,
+      "step": 865
+    },
+    {
+      "epoch": 0.13613944074357917,
+      "grad_norm": 0.22028383612632751,
+      "learning_rate": 4.944397061386794e-05,
+      "loss": 1.2494,
+      "step": 866
+    },
+    {
+      "epoch": 0.1362966456405117,
+      "grad_norm": 0.2354927659034729,
+      "learning_rate": 4.944267405680164e-05,
+      "loss": 1.1469,
+      "step": 867
+    },
+    {
+      "epoch": 0.13645385053744424,
+      "grad_norm": 0.28941988945007324,
+      "learning_rate": 4.944137600686981e-05,
+      "loss": 1.1678,
+      "step": 868
+    },
+    {
+      "epoch": 0.1366110554343768,
+      "grad_norm": 0.2538214325904846,
+      "learning_rate": 4.944007646415172e-05,
+      "loss": 1.2636,
+      "step": 869
+    },
+    {
+      "epoch": 0.13676826033130932,
+      "grad_norm": 0.3719157576560974,
+      "learning_rate": 4.943877542872676e-05,
+      "loss": 1.2901,
+      "step": 870
+    },
+    {
+      "epoch": 0.13692546522824187,
+      "grad_norm": 0.2994091212749481,
+      "learning_rate": 4.943747290067438e-05,
+      "loss": 1.2209,
+      "step": 871
+    },
+    {
+      "epoch": 0.1370826701251744,
+      "grad_norm": 0.23586580157279968,
+      "learning_rate": 4.9436168880074115e-05,
+      "loss": 1.2989,
+      "step": 872
+    },
+    {
+      "epoch": 0.13723987502210694,
+      "grad_norm": 0.193126380443573,
+      "learning_rate": 4.943486336700564e-05,
+      "loss": 1.204,
+      "step": 873
+    },
+    {
+      "epoch": 0.1373970799190395,
+      "grad_norm": 0.18505080044269562,
+      "learning_rate": 4.943355636154868e-05,
+      "loss": 1.3247,
+      "step": 874
+    },
+    {
+      "epoch": 0.137554284815972,
+      "grad_norm": 0.2586881220340729,
+      "learning_rate": 4.9432247863783064e-05,
+      "loss": 1.3315,
+      "step": 875
+    },
+    {
+      "epoch": 0.13771148971290456,
+      "grad_norm": 0.2904506027698517,
+      "learning_rate": 4.943093787378871e-05,
+      "loss": 1.2593,
+      "step": 876
+    },
+    {
+      "epoch": 0.13786869460983708,
+      "grad_norm": 0.2971174120903015,
+      "learning_rate": 4.9429626391645615e-05,
+      "loss": 1.2241,
+      "step": 877
+    },
+    {
+      "epoch": 0.13802589950676963,
+      "grad_norm": 0.42521703243255615,
+      "learning_rate": 4.9428313417433894e-05,
+      "loss": 1.2638,
+      "step": 878
+    },
+    {
+      "epoch": 0.13818310440370218,
+      "grad_norm": 0.2515777349472046,
+      "learning_rate": 4.9426998951233735e-05,
+      "loss": 1.3111,
+      "step": 879
+    },
+    {
+      "epoch": 0.1383403093006347,
+      "grad_norm": 0.25959545373916626,
+      "learning_rate": 4.942568299312541e-05,
+      "loss": 1.2505,
+      "step": 880
+    },
+    {
+      "epoch": 0.13849751419756726,
+      "grad_norm": 0.28090932965278625,
+      "learning_rate": 4.942436554318931e-05,
+      "loss": 1.1604,
+      "step": 881
+    },
+    {
+      "epoch": 0.1386547190944998,
+      "grad_norm": 0.21833541989326477,
+      "learning_rate": 4.942304660150588e-05,
+      "loss": 1.2246,
+      "step": 882
+    },
+    {
+      "epoch": 0.13881192399143233,
+      "grad_norm": 0.26167765259742737,
+      "learning_rate": 4.9421726168155704e-05,
+      "loss": 1.2399,
+      "step": 883
+    },
+    {
+      "epoch": 0.13896912888836488,
+      "grad_norm": 0.23778817057609558,
+      "learning_rate": 4.9420404243219395e-05,
+      "loss": 1.2692,
+      "step": 884
+    },
+    {
+      "epoch": 0.1391263337852974,
+      "grad_norm": 0.43253734707832336,
+      "learning_rate": 4.941908082677773e-05,
+      "loss": 1.2302,
+      "step": 885
+    },
+    {
+      "epoch": 0.13928353868222995,
+      "grad_norm": 0.2448786050081253,
+      "learning_rate": 4.94177559189115e-05,
+      "loss": 1.3163,
+      "step": 886
+    },
+    {
+      "epoch": 0.1394407435791625,
+      "grad_norm": 0.24711786210536957,
+      "learning_rate": 4.941642951970165e-05,
+      "loss": 1.2756,
+      "step": 887
+    },
+    {
+      "epoch": 0.13959794847609502,
+      "grad_norm": 0.22932004928588867,
+      "learning_rate": 4.941510162922917e-05,
+      "loss": 1.3087,
+      "step": 888
+    },
+    {
+      "epoch": 0.13975515337302757,
+      "grad_norm": 0.24999158084392548,
+      "learning_rate": 4.941377224757518e-05,
+      "loss": 1.3328,
+      "step": 889
+    },
+    {
+      "epoch": 0.1399123582699601,
+      "grad_norm": 0.21222981810569763,
+      "learning_rate": 4.941244137482088e-05,
+      "loss": 1.3177,
+      "step": 890
+    },
+    {
+      "epoch": 0.14006956316689265,
+      "grad_norm": 0.22691656649112701,
+      "learning_rate": 4.941110901104754e-05,
+      "loss": 1.2937,
+      "step": 891
+    },
+    {
+      "epoch": 0.1402267680638252,
+      "grad_norm": 0.3120933771133423,
+      "learning_rate": 4.940977515633653e-05,
+      "loss": 1.1604,
+      "step": 892
+    },
+    {
+      "epoch": 0.14038397296075772,
+      "grad_norm": 0.24279998242855072,
+      "learning_rate": 4.940843981076934e-05,
+      "loss": 1.3234,
+      "step": 893
+    },
+    {
+      "epoch": 0.14054117785769027,
+      "grad_norm": 0.25406959652900696,
+      "learning_rate": 4.940710297442751e-05,
+      "loss": 1.3216,
+      "step": 894
+    },
+    {
+      "epoch": 0.14069838275462282,
+      "grad_norm": 0.29678472876548767,
+      "learning_rate": 4.940576464739269e-05,
+      "loss": 1.2706,
+      "step": 895
+    },
+    {
+      "epoch": 0.14085558765155534,
+      "grad_norm": 0.25185081362724304,
+      "learning_rate": 4.9404424829746634e-05,
+      "loss": 1.2456,
+      "step": 896
+    },
+    {
+      "epoch": 0.1410127925484879,
+      "grad_norm": 0.2171952873468399,
+      "learning_rate": 4.940308352157115e-05,
+      "loss": 1.2943,
+      "step": 897
+    },
+    {
+      "epoch": 0.14116999744542041,
+      "grad_norm": 0.21498677134513855,
+      "learning_rate": 4.940174072294818e-05,
+      "loss": 1.3466,
+      "step": 898
+    },
+    {
+      "epoch": 0.14132720234235296,
+      "grad_norm": 0.2881999611854553,
+      "learning_rate": 4.940039643395972e-05,
+      "loss": 1.2322,
+      "step": 899
+    },
+    {
+      "epoch": 0.14148440723928551,
+      "grad_norm": 0.2709384858608246,
+      "learning_rate": 4.939905065468789e-05,
+      "loss": 1.2228,
+      "step": 900
+    },
+    {
+      "epoch": 0.14164161213621804,
+      "grad_norm": 0.2723088562488556,
+      "learning_rate": 4.9397703385214875e-05,
+      "loss": 1.1937,
+      "step": 901
+    },
+    {
+      "epoch": 0.1417988170331506,
+      "grad_norm": 0.4296363294124603,
+      "learning_rate": 4.939635462562297e-05,
+      "loss": 1.2043,
+      "step": 902
+    },
+    {
+      "epoch": 0.1419560219300831,
+      "grad_norm": 0.3255182206630707,
+      "learning_rate": 4.939500437599454e-05,
+      "loss": 1.1563,
+      "step": 903
+    },
+    {
+      "epoch": 0.14211322682701566,
+      "grad_norm": 0.33772897720336914,
+      "learning_rate": 4.939365263641206e-05,
+      "loss": 1.3019,
+      "step": 904
+    },
+    {
+      "epoch": 0.1422704317239482,
+      "grad_norm": 0.18991219997406006,
+      "learning_rate": 4.93922994069581e-05,
+      "loss": 1.3575,
+      "step": 905
+    },
+    {
+      "epoch": 0.14242763662088073,
+      "grad_norm": 0.23950403928756714,
+      "learning_rate": 4.939094468771529e-05,
+      "loss": 1.2512,
+      "step": 906
+    },
+    {
+      "epoch": 0.14258484151781328,
+      "grad_norm": 0.29783302545547485,
+      "learning_rate": 4.938958847876637e-05,
+      "loss": 1.3033,
+      "step": 907
+    },
+    {
+      "epoch": 0.14274204641474583,
+      "grad_norm": 0.3168744742870331,
+      "learning_rate": 4.93882307801942e-05,
+      "loss": 1.1934,
+      "step": 908
+    },
+    {
+      "epoch": 0.14289925131167835,
+      "grad_norm": 0.22578391432762146,
+      "learning_rate": 4.9386871592081675e-05,
+      "loss": 1.3307,
+      "step": 909
+    },
+    {
+      "epoch": 0.1430564562086109,
+      "grad_norm": 0.32671108841896057,
+      "learning_rate": 4.9385510914511824e-05,
+      "loss": 1.2436,
+      "step": 910
+    },
+    {
+      "epoch": 0.14321366110554343,
+      "grad_norm": 0.2524665296077728,
+      "learning_rate": 4.938414874756774e-05,
+      "loss": 1.2611,
+      "step": 911
+    },
+    {
+      "epoch": 0.14337086600247598,
+      "grad_norm": 0.3576960563659668,
+      "learning_rate": 4.9382785091332625e-05,
+      "loss": 1.3721,
+      "step": 912
+    },
+    {
+      "epoch": 0.14352807089940853,
+      "grad_norm": 0.2915900945663452,
+      "learning_rate": 4.9381419945889776e-05,
+      "loss": 1.3539,
+      "step": 913
+    },
+    {
+      "epoch": 0.14368527579634105,
+      "grad_norm": 0.3168608844280243,
+      "learning_rate": 4.938005331132256e-05,
+      "loss": 1.224,
+      "step": 914
+    },
+    {
+      "epoch": 0.1438424806932736,
+      "grad_norm": 0.24886426329612732,
+      "learning_rate": 4.937868518771445e-05,
+      "loss": 1.2299,
+      "step": 915
+    },
+    {
+      "epoch": 0.14399968559020612,
+      "grad_norm": 0.26588642597198486,
+      "learning_rate": 4.9377315575149e-05,
+      "loss": 1.1947,
+      "step": 916
+    },
+    {
+      "epoch": 0.14415689048713867,
+      "grad_norm": 0.28032201528549194,
+      "learning_rate": 4.937594447370986e-05,
+      "loss": 1.3756,
+      "step": 917
+    },
+    {
+      "epoch": 0.14431409538407122,
+      "grad_norm": 0.3017072081565857,
+      "learning_rate": 4.937457188348078e-05,
+      "loss": 1.2723,
+      "step": 918
+    },
+    {
+      "epoch": 0.14447130028100374,
+      "grad_norm": 0.2926197648048401,
+      "learning_rate": 4.937319780454559e-05,
+      "loss": 1.2716,
+      "step": 919
+    },
+    {
+      "epoch": 0.1446285051779363,
+      "grad_norm": 0.24066713452339172,
+      "learning_rate": 4.937182223698821e-05,
+      "loss": 1.2828,
+      "step": 920
+    },
+    {
+      "epoch": 0.14478571007486885,
+      "grad_norm": 0.30001577734947205,
+      "learning_rate": 4.937044518089266e-05,
+      "loss": 1.2407,
+      "step": 921
+    },
+    {
+      "epoch": 0.14494291497180137,
+      "grad_norm": 0.25927406549453735,
+      "learning_rate": 4.9369066636343044e-05,
+      "loss": 1.3004,
+      "step": 922
+    },
+    {
+      "epoch": 0.14510011986873392,
+      "grad_norm": 0.2542930543422699,
+      "learning_rate": 4.936768660342355e-05,
+      "loss": 1.3312,
+      "step": 923
+    },
+    {
+      "epoch": 0.14525732476566644,
+      "grad_norm": 0.25233832001686096,
+      "learning_rate": 4.936630508221847e-05,
+      "loss": 1.1879,
+      "step": 924
+    },
+    {
+      "epoch": 0.145414529662599,
+      "grad_norm": 0.22136953473091125,
+      "learning_rate": 4.9364922072812185e-05,
+      "loss": 1.2649,
+      "step": 925
+    },
+    {
+      "epoch": 0.14557173455953154,
+      "grad_norm": 0.21759863197803497,
+      "learning_rate": 4.936353757528916e-05,
+      "loss": 1.2467,
+      "step": 926
+    },
+    {
+      "epoch": 0.14572893945646406,
+      "grad_norm": 0.27614825963974,
+      "learning_rate": 4.936215158973396e-05,
+      "loss": 1.1901,
+      "step": 927
+    },
+    {
+      "epoch": 0.1458861443533966,
+      "grad_norm": 0.2502923309803009,
+      "learning_rate": 4.936076411623124e-05,
+      "loss": 1.3358,
+      "step": 928
+    },
+    {
+      "epoch": 0.14604334925032914,
+      "grad_norm": 0.2419285923242569,
+      "learning_rate": 4.935937515486573e-05,
+      "loss": 1.24,
+      "step": 929
+    },
+    {
+      "epoch": 0.14620055414726169,
+      "grad_norm": 0.35315272212028503,
+      "learning_rate": 4.935798470572226e-05,
+      "loss": 1.2452,
+      "step": 930
+    },
+    {
+      "epoch": 0.14635775904419424,
+      "grad_norm": 0.28915464878082275,
+      "learning_rate": 4.935659276888577e-05,
+      "loss": 1.3369,
+      "step": 931
+    },
+    {
+      "epoch": 0.14651496394112676,
+      "grad_norm": 0.23898139595985413,
+      "learning_rate": 4.9355199344441254e-05,
+      "loss": 1.2328,
+      "step": 932
+    },
+    {
+      "epoch": 0.1466721688380593,
+      "grad_norm": 0.25197896361351013,
+      "learning_rate": 4.935380443247384e-05,
+      "loss": 1.2826,
+      "step": 933
+    },
+    {
+      "epoch": 0.14682937373499186,
+      "grad_norm": 0.26547369360923767,
+      "learning_rate": 4.9352408033068695e-05,
+      "loss": 1.2284,
+      "step": 934
+    },
+    {
+      "epoch": 0.14698657863192438,
+      "grad_norm": 0.22031289339065552,
+      "learning_rate": 4.935101014631114e-05,
+      "loss": 1.2918,
+      "step": 935
+    },
+    {
+      "epoch": 0.14714378352885693,
+      "grad_norm": 0.2603214979171753,
+      "learning_rate": 4.9349610772286525e-05,
+      "loss": 1.1767,
+      "step": 936
+    },
+    {
+      "epoch": 0.14730098842578945,
+      "grad_norm": 0.29469192028045654,
+      "learning_rate": 4.934820991108032e-05,
+      "loss": 1.2845,
+      "step": 937
+    },
+    {
+      "epoch": 0.147458193322722,
+      "grad_norm": 0.30825692415237427,
+      "learning_rate": 4.934680756277811e-05,
+      "loss": 1.1999,
+      "step": 938
+    },
+    {
+      "epoch": 0.14761539821965455,
+      "grad_norm": 0.25342094898223877,
+      "learning_rate": 4.934540372746552e-05,
+      "loss": 1.2285,
+      "step": 939
+    },
+    {
+      "epoch": 0.14777260311658708,
+      "grad_norm": 0.26036733388900757,
+      "learning_rate": 4.9343998405228295e-05,
+      "loss": 1.2367,
+      "step": 940
+    },
+    {
+      "epoch": 0.14792980801351963,
+      "grad_norm": 0.27401411533355713,
+      "learning_rate": 4.934259159615228e-05,
+      "loss": 1.1985,
+      "step": 941
+    },
+    {
+      "epoch": 0.14808701291045215,
+      "grad_norm": 0.23039095103740692,
+      "learning_rate": 4.934118330032338e-05,
+      "loss": 1.2649,
+      "step": 942
+    },
+    {
+      "epoch": 0.1482442178073847,
+      "grad_norm": 0.29547953605651855,
+      "learning_rate": 4.933977351782761e-05,
+      "loss": 1.1345,
+      "step": 943
+    },
+    {
+      "epoch": 0.14840142270431725,
+      "grad_norm": 0.22598884999752045,
+      "learning_rate": 4.933836224875109e-05,
+      "loss": 1.2965,
+      "step": 944
+    },
+    {
+      "epoch": 0.14855862760124977,
+      "grad_norm": 0.31008240580558777,
+      "learning_rate": 4.9336949493180006e-05,
+      "loss": 1.1144,
+      "step": 945
+    },
+    {
+      "epoch": 0.14871583249818232,
+      "grad_norm": 0.28397658467292786,
+      "learning_rate": 4.9335535251200636e-05,
+      "loss": 1.266,
+      "step": 946
+    },
+    {
+      "epoch": 0.14887303739511487,
+      "grad_norm": 0.2284776121377945,
+      "learning_rate": 4.933411952289937e-05,
+      "loss": 1.2164,
+      "step": 947
+    },
+    {
+      "epoch": 0.1490302422920474,
+      "grad_norm": 0.2107551246881485,
+      "learning_rate": 4.9332702308362665e-05,
+      "loss": 1.2719,
+      "step": 948
+    },
+    {
+      "epoch": 0.14918744718897994,
+      "grad_norm": 0.26652616262435913,
+      "learning_rate": 4.933128360767709e-05,
+      "loss": 1.2304,
+      "step": 949
+    },
+    {
+      "epoch": 0.14934465208591247,
+      "grad_norm": 0.22624680399894714,
+      "learning_rate": 4.932986342092928e-05,
+      "loss": 1.2999,
+      "step": 950
+    },
+    {
+      "epoch": 0.14950185698284502,
+      "grad_norm": 0.20410288870334625,
+      "learning_rate": 4.932844174820598e-05,
+      "loss": 1.2269,
+      "step": 951
+    },
+    {
+      "epoch": 0.14965906187977757,
+      "grad_norm": 0.24987919628620148,
+      "learning_rate": 4.932701858959403e-05,
+      "loss": 1.3042,
+      "step": 952
+    },
+    {
+      "epoch": 0.1498162667767101,
+      "grad_norm": 0.191947340965271,
+      "learning_rate": 4.932559394518033e-05,
+      "loss": 1.2803,
+      "step": 953
+    },
+    {
+      "epoch": 0.14997347167364264,
+      "grad_norm": 0.3396085798740387,
+      "learning_rate": 4.932416781505191e-05,
+      "loss": 1.2014,
+      "step": 954
+    },
+    {
+      "epoch": 0.15013067657057516,
+      "grad_norm": 0.22375932335853577,
+      "learning_rate": 4.932274019929587e-05,
+      "loss": 1.242,
+      "step": 955
+    },
+    {
+      "epoch": 0.1502878814675077,
+      "grad_norm": 0.281097412109375,
+      "learning_rate": 4.93213110979994e-05,
+      "loss": 1.2742,
+      "step": 956
+    },
+    {
+      "epoch": 0.15044508636444026,
+      "grad_norm": 0.24049919843673706,
+      "learning_rate": 4.931988051124979e-05,
+      "loss": 1.3166,
+      "step": 957
+    },
+    {
+      "epoch": 0.15060229126137278,
+      "grad_norm": 0.24433936178684235,
+      "learning_rate": 4.93184484391344e-05,
+      "loss": 1.2933,
+      "step": 958
+    },
+    {
+      "epoch": 0.15075949615830533,
+      "grad_norm": 0.3671477138996124,
+      "learning_rate": 4.9317014881740706e-05,
+      "loss": 1.1731,
+      "step": 959
+    },
+    {
+      "epoch": 0.15091670105523788,
+      "grad_norm": 0.22575189173221588,
+      "learning_rate": 4.931557983915627e-05,
+      "loss": 1.2509,
+      "step": 960
+    },
+    {
+      "epoch": 0.15091670105523788,
+      "eval_loss": 1.2455451488494873,
+      "eval_runtime": 2308.563,
+      "eval_samples_per_second": 4.01,
+      "eval_steps_per_second": 2.005,
+      "step": 960
+    },
+    {
+      "epoch": 0.1510739059521704,
+      "grad_norm": 0.25838157534599304,
+      "learning_rate": 4.931414331146873e-05,
+      "loss": 1.3554,
+      "step": 961
+    },
+    {
+      "epoch": 0.15123111084910296,
+      "grad_norm": 0.3163435459136963,
+      "learning_rate": 4.931270529876583e-05,
+      "loss": 1.3133,
+      "step": 962
+    },
+    {
+      "epoch": 0.15138831574603548,
+      "grad_norm": 0.30024880170822144,
+      "learning_rate": 4.9311265801135384e-05,
+      "loss": 1.2303,
+      "step": 963
+    },
+    {
+      "epoch": 0.15154552064296803,
+      "grad_norm": 0.3424816429615021,
+      "learning_rate": 4.9309824818665325e-05,
+      "loss": 1.1929,
+      "step": 964
+    },
+    {
+      "epoch": 0.15170272553990058,
+      "grad_norm": 0.27401861548423767,
+      "learning_rate": 4.930838235144366e-05,
+      "loss": 1.2198,
+      "step": 965
+    },
+    {
+      "epoch": 0.1518599304368331,
+      "grad_norm": 0.24110247194766998,
+      "learning_rate": 4.930693839955848e-05,
+      "loss": 1.2381,
+      "step": 966
+    },
+    {
+      "epoch": 0.15201713533376565,
+      "grad_norm": 0.159100741147995,
+      "learning_rate": 4.9305492963098e-05,
+      "loss": 1.2666,
+      "step": 967
+    },
+    {
+      "epoch": 0.15217434023069817,
+      "grad_norm": 0.23810729384422302,
+      "learning_rate": 4.9304046042150474e-05,
+      "loss": 1.2592,
+      "step": 968
+    },
+    {
+      "epoch": 0.15233154512763072,
+      "grad_norm": 0.19887159764766693,
+      "learning_rate": 4.930259763680429e-05,
+      "loss": 1.3179,
+      "step": 969
+    },
+    {
+      "epoch": 0.15248875002456327,
+      "grad_norm": 0.22060149908065796,
+      "learning_rate": 4.930114774714791e-05,
+      "loss": 1.2712,
+      "step": 970
+    },
+    {
+      "epoch": 0.1526459549214958,
+      "grad_norm": 0.2443406730890274,
+      "learning_rate": 4.929969637326989e-05,
+      "loss": 1.2021,
+      "step": 971
+    },
+    {
+      "epoch": 0.15280315981842835,
+      "grad_norm": 0.2185499370098114,
+      "learning_rate": 4.9298243515258855e-05,
+      "loss": 1.3017,
+      "step": 972
+    },
+    {
+      "epoch": 0.1529603647153609,
+      "grad_norm": 0.23159849643707275,
+      "learning_rate": 4.929678917320357e-05,
+      "loss": 1.2122,
+      "step": 973
+    },
+    {
+      "epoch": 0.15311756961229342,
+      "grad_norm": 0.3281627595424652,
+      "learning_rate": 4.929533334719284e-05,
+      "loss": 1.1151,
+      "step": 974
+    },
+    {
+      "epoch": 0.15327477450922597,
+      "grad_norm": 0.24456332623958588,
+      "learning_rate": 4.929387603731558e-05,
+      "loss": 1.2107,
+      "step": 975
+    },
+    {
+      "epoch": 0.1534319794061585,
+      "grad_norm": 0.3623602092266083,
+      "learning_rate": 4.9292417243660814e-05,
+      "loss": 1.3041,
+      "step": 976
+    },
+    {
+      "epoch": 0.15358918430309104,
+      "grad_norm": 0.2319325953722,
+      "learning_rate": 4.929095696631763e-05,
+      "loss": 1.2331,
+      "step": 977
+    },
+    {
+      "epoch": 0.1537463892000236,
+      "grad_norm": 0.324660986661911,
+      "learning_rate": 4.92894952053752e-05,
+      "loss": 1.2511,
+      "step": 978
+    },
+    {
+      "epoch": 0.15390359409695611,
+      "grad_norm": 0.23866824805736542,
+      "learning_rate": 4.9288031960922834e-05,
+      "loss": 1.2709,
+      "step": 979
+    },
+    {
+      "epoch": 0.15406079899388866,
+      "grad_norm": 0.21922123432159424,
+      "learning_rate": 4.928656723304989e-05,
+      "loss": 1.3108,
+      "step": 980
+    },
+    {
+      "epoch": 0.1542180038908212,
+      "grad_norm": 0.2919687032699585,
+      "learning_rate": 4.92851010218458e-05,
+      "loss": 1.0465,
+      "step": 981
+    },
+    {
+      "epoch": 0.15437520878775374,
+      "grad_norm": 0.26380711793899536,
+      "learning_rate": 4.9283633327400156e-05,
+      "loss": 1.2066,
+      "step": 982
+    },
+    {
+      "epoch": 0.1545324136846863,
+      "grad_norm": 0.2607104182243347,
+      "learning_rate": 4.9282164149802576e-05,
+      "loss": 1.1227,
+      "step": 983
+    },
+    {
+      "epoch": 0.1546896185816188,
+      "grad_norm": 0.32302606105804443,
+      "learning_rate": 4.92806934891428e-05,
+      "loss": 1.2066,
+      "step": 984
+    },
+    {
+      "epoch": 0.15484682347855136,
+      "grad_norm": 0.28476232290267944,
+      "learning_rate": 4.927922134551065e-05,
+      "loss": 1.1447,
+      "step": 985
+    },
+    {
+      "epoch": 0.1550040283754839,
+      "grad_norm": 0.253738671541214,
+      "learning_rate": 4.9277747718996036e-05,
+      "loss": 1.2438,
+      "step": 986
+    },
+    {
+      "epoch": 0.15516123327241643,
+      "grad_norm": 0.29659610986709595,
+      "learning_rate": 4.927627260968896e-05,
+      "loss": 1.1946,
+      "step": 987
+    },
+    {
+      "epoch": 0.15531843816934898,
+      "grad_norm": 0.27436473965644836,
+      "learning_rate": 4.927479601767952e-05,
+      "loss": 1.2783,
+      "step": 988
+    },
+    {
+      "epoch": 0.1554756430662815,
+      "grad_norm": 0.21257497370243073,
+      "learning_rate": 4.9273317943057896e-05,
+      "loss": 1.204,
+      "step": 989
+    },
+    {
+      "epoch": 0.15563284796321405,
+      "grad_norm": 0.2546120584011078,
+      "learning_rate": 4.927183838591437e-05,
+      "loss": 1.2224,
+      "step": 990
+    },
+    {
+      "epoch": 0.1557900528601466,
+      "grad_norm": 0.2336018979549408,
+      "learning_rate": 4.92703573463393e-05,
+      "loss": 1.2903,
+      "step": 991
+    },
+    {
+      "epoch": 0.15594725775707913,
+      "grad_norm": 0.2048567533493042,
+      "learning_rate": 4.926887482442315e-05,
+      "loss": 1.2529,
+      "step": 992
+    },
+    {
+      "epoch": 0.15610446265401168,
+      "grad_norm": 0.20950450003147125,
+      "learning_rate": 4.926739082025646e-05,
+      "loss": 1.4247,
+      "step": 993
+    },
+    {
+      "epoch": 0.1562616675509442,
+      "grad_norm": 0.22965438663959503,
+      "learning_rate": 4.926590533392987e-05,
+      "loss": 1.2573,
+      "step": 994
+    },
+    {
+      "epoch": 0.15641887244787675,
+      "grad_norm": 0.25654879212379456,
+      "learning_rate": 4.9264418365534105e-05,
+      "loss": 1.2115,
+      "step": 995
+    },
+    {
+      "epoch": 0.1565760773448093,
+      "grad_norm": 0.26419493556022644,
+      "learning_rate": 4.9262929915159995e-05,
+      "loss": 1.2994,
+      "step": 996
+    },
+    {
+      "epoch": 0.15673328224174182,
+      "grad_norm": 0.3215327560901642,
+      "learning_rate": 4.926143998289843e-05,
+      "loss": 1.2268,
+      "step": 997
+    },
+    {
+      "epoch": 0.15689048713867437,
+      "grad_norm": 0.3190024495124817,
+      "learning_rate": 4.925994856884042e-05,
+      "loss": 1.2747,
+      "step": 998
+    },
+    {
+      "epoch": 0.15704769203560692,
+      "grad_norm": 0.26742681860923767,
+      "learning_rate": 4.9258455673077065e-05,
+      "loss": 1.2791,
+      "step": 999
+    },
+    {
+      "epoch": 0.15720489693253945,
+      "grad_norm": 0.21026930212974548,
+      "learning_rate": 4.925696129569953e-05,
+      "loss": 1.3694,
+      "step": 1000
+    },
+    {
+      "epoch": 0.157362101829472,
+      "grad_norm": 0.2838928699493408,
+      "learning_rate": 4.925546543679909e-05,
+      "loss": 1.2589,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15751930672640452,
+      "grad_norm": 0.3007844090461731,
+      "learning_rate": 4.9253968096467104e-05,
+      "loss": 1.2488,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15767651162333707,
+      "grad_norm": 0.2473154515028,
+      "learning_rate": 4.925246927479503e-05,
+      "loss": 1.2288,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15783371652026962,
+      "grad_norm": 0.2360457479953766,
+      "learning_rate": 4.925096897187441e-05,
+      "loss": 1.3148,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15799092141720214,
+      "grad_norm": 0.3671962320804596,
+      "learning_rate": 4.924946718779687e-05,
+      "loss": 1.1999,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1581481263141347,
+      "grad_norm": 0.28391456604003906,
+      "learning_rate": 4.924796392265414e-05,
+      "loss": 1.2296,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1583053312110672,
+      "grad_norm": 0.2550790011882782,
+      "learning_rate": 4.924645917653802e-05,
+      "loss": 1.1923,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15846253610799976,
+      "grad_norm": 0.301740825176239,
+      "learning_rate": 4.924495294954044e-05,
+      "loss": 1.2363,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1586197410049323,
+      "grad_norm": 0.2608097195625305,
+      "learning_rate": 4.9243445241753374e-05,
+      "loss": 1.298,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15877694590186484,
+      "grad_norm": 0.31052571535110474,
+      "learning_rate": 4.924193605326891e-05,
+      "loss": 1.3037,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15893415079879739,
+      "grad_norm": 0.2420274317264557,
+      "learning_rate": 4.924042538417923e-05,
+      "loss": 1.2541,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15909135569572994,
+      "grad_norm": 0.21056459844112396,
+      "learning_rate": 4.92389132345766e-05,
+      "loss": 1.3229,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15924856059266246,
+      "grad_norm": 0.3283076286315918,
+      "learning_rate": 4.923739960455337e-05,
+      "loss": 1.1036,
+      "step": 1013
+    },
+    {
+      "epoch": 0.159405765489595,
+      "grad_norm": 0.2461448460817337,
+      "learning_rate": 4.9235884494201987e-05,
+      "loss": 1.2503,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15956297038652753,
+      "grad_norm": 0.2919829189777374,
+      "learning_rate": 4.923436790361499e-05,
+      "loss": 1.2108,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15972017528346008,
+      "grad_norm": 0.22029368579387665,
+      "learning_rate": 4.923284983288501e-05,
+      "loss": 1.2369,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15987738018039263,
+      "grad_norm": 0.27863630652427673,
+      "learning_rate": 4.9231330282104756e-05,
+      "loss": 1.2555,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16003458507732515,
+      "grad_norm": 0.28123825788497925,
+      "learning_rate": 4.9229809251367055e-05,
+      "loss": 1.1519,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1601917899742577,
+      "grad_norm": 0.231231227517128,
+      "learning_rate": 4.922828674076478e-05,
+      "loss": 1.3488,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16034899487119023,
+      "grad_norm": 0.2704828977584839,
+      "learning_rate": 4.9226762750390944e-05,
+      "loss": 1.1102,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16050619976812278,
+      "grad_norm": 0.17970627546310425,
+      "learning_rate": 4.922523728033861e-05,
+      "loss": 1.2237,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16066340466505533,
+      "grad_norm": 0.22386445105075836,
+      "learning_rate": 4.9223710330700956e-05,
+      "loss": 1.2564,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16082060956198785,
+      "grad_norm": 0.21347114443778992,
+      "learning_rate": 4.922218190157124e-05,
+      "loss": 1.2433,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1609778144589204,
+      "grad_norm": 0.23873676359653473,
+      "learning_rate": 4.9220651993042813e-05,
+      "loss": 1.2018,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16113501935585295,
+      "grad_norm": 0.26954975724220276,
+      "learning_rate": 4.921912060520912e-05,
+      "loss": 1.2118,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16129222425278547,
+      "grad_norm": 0.3023718595504761,
+      "learning_rate": 4.9217587738163686e-05,
+      "loss": 1.2717,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16144942914971802,
+      "grad_norm": 0.31107062101364136,
+      "learning_rate": 4.921605339200013e-05,
+      "loss": 1.2017,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16160663404665054,
+      "grad_norm": 0.2795855402946472,
+      "learning_rate": 4.921451756681217e-05,
+      "loss": 1.3492,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1617638389435831,
+      "grad_norm": 0.24515774846076965,
+      "learning_rate": 4.921298026269361e-05,
+      "loss": 1.2763,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16192104384051564,
+      "grad_norm": 0.2603877782821655,
+      "learning_rate": 4.921144147973834e-05,
+      "loss": 1.3196,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16207824873744817,
+      "grad_norm": 0.265697181224823,
+      "learning_rate": 4.9209901218040335e-05,
+      "loss": 1.3067,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16223545363438072,
+      "grad_norm": 0.2554314136505127,
+      "learning_rate": 4.9208359477693686e-05,
+      "loss": 1.2347,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16239265853131324,
+      "grad_norm": 0.3272973597049713,
+      "learning_rate": 4.920681625879254e-05,
+      "loss": 1.2104,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1625498634282458,
+      "grad_norm": 0.21259547770023346,
+      "learning_rate": 4.9205271561431166e-05,
+      "loss": 1.2857,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16270706832517834,
+      "grad_norm": 0.2505529820919037,
+      "learning_rate": 4.92037253857039e-05,
+      "loss": 1.1988,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16286427322211086,
+      "grad_norm": 0.2369750738143921,
+      "learning_rate": 4.920217773170517e-05,
+      "loss": 1.2384,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1630214781190434,
+      "grad_norm": 0.26577889919281006,
+      "learning_rate": 4.920062859952951e-05,
+      "loss": 1.1521,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16317868301597593,
+      "grad_norm": 0.2224215418100357,
+      "learning_rate": 4.919907798927153e-05,
+      "loss": 1.3065,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16333588791290848,
+      "grad_norm": 0.1882622092962265,
+      "learning_rate": 4.9197525901025944e-05,
+      "loss": 1.2472,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16349309280984103,
+      "grad_norm": 0.25392916798591614,
+      "learning_rate": 4.919597233488754e-05,
+      "loss": 1.3387,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16365029770677356,
+      "grad_norm": 0.3185995817184448,
+      "learning_rate": 4.91944172909512e-05,
+      "loss": 1.161,
+      "step": 1041
+    },
+    {
+      "epoch": 0.1638075026037061,
+      "grad_norm": 0.26924118399620056,
+      "learning_rate": 4.919286076931191e-05,
+      "loss": 1.0659,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16396470750063866,
+      "grad_norm": 0.2224770486354828,
+      "learning_rate": 4.919130277006473e-05,
+      "loss": 1.2303,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16412191239757118,
+      "grad_norm": 0.24008037149906158,
+      "learning_rate": 4.918974329330482e-05,
+      "loss": 1.2762,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16427911729450373,
+      "grad_norm": 0.2728358209133148,
+      "learning_rate": 4.918818233912742e-05,
+      "loss": 1.1582,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16443632219143625,
+      "grad_norm": 0.24911800026893616,
+      "learning_rate": 4.918661990762788e-05,
+      "loss": 1.2155,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1645935270883688,
+      "grad_norm": 0.2444472759962082,
+      "learning_rate": 4.918505599890162e-05,
+      "loss": 1.2838,
+      "step": 1047
+    },
+    {
+      "epoch": 0.16475073198530135,
+      "grad_norm": 0.2379113882780075,
+      "learning_rate": 4.918349061304416e-05,
+      "loss": 1.3043,
+      "step": 1048
+    },
+    {
+      "epoch": 0.16490793688223387,
+      "grad_norm": 0.3085183799266815,
+      "learning_rate": 4.9181923750151095e-05,
+      "loss": 1.2568,
+      "step": 1049
+    },
+    {
+      "epoch": 0.16506514177916642,
+      "grad_norm": 0.2629674971103668,
+      "learning_rate": 4.918035541031814e-05,
+      "loss": 1.2171,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16522234667609895,
+      "grad_norm": 0.2707282602787018,
+      "learning_rate": 4.917878559364107e-05,
+      "loss": 1.1597,
+      "step": 1051
+    },
+    {
+      "epoch": 0.1653795515730315,
+      "grad_norm": 0.27305370569229126,
+      "learning_rate": 4.9177214300215784e-05,
+      "loss": 1.36,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16553675646996405,
+      "grad_norm": 0.20558474957942963,
+      "learning_rate": 4.9175641530138226e-05,
+      "loss": 1.2225,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16569396136689657,
+      "grad_norm": 0.23680521547794342,
+      "learning_rate": 4.917406728350448e-05,
+      "loss": 1.2149,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16585116626382912,
+      "grad_norm": 0.2101297229528427,
+      "learning_rate": 4.917249156041066e-05,
+      "loss": 1.2313,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16600837116076167,
+      "grad_norm": 0.2601447105407715,
+      "learning_rate": 4.917091436095304e-05,
+      "loss": 1.1907,
+      "step": 1056
+    },
+    {
+      "epoch": 0.1661655760576942,
+      "grad_norm": 0.2113189995288849,
+      "learning_rate": 4.916933568522793e-05,
+      "loss": 1.2852,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16632278095462674,
+      "grad_norm": 0.21135227382183075,
+      "learning_rate": 4.916775553333176e-05,
+      "loss": 1.2852,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16647998585155926,
+      "grad_norm": 0.2743116617202759,
+      "learning_rate": 4.916617390536102e-05,
+      "loss": 1.2032,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16663719074849181,
+      "grad_norm": 0.2520056664943695,
+      "learning_rate": 4.916459080141234e-05,
+      "loss": 1.3038,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16679439564542436,
+      "grad_norm": 0.21614307165145874,
+      "learning_rate": 4.916300622158239e-05,
+      "loss": 1.2216,
+      "step": 1061
+    },
+    {
+      "epoch": 0.1669516005423569,
+      "grad_norm": 0.28615444898605347,
+      "learning_rate": 4.9161420165967956e-05,
+      "loss": 1.2162,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16710880543928944,
+      "grad_norm": 0.33522137999534607,
+      "learning_rate": 4.91598326346659e-05,
+      "loss": 1.1458,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16726601033622196,
+      "grad_norm": 0.30597957968711853,
+      "learning_rate": 4.9158243627773194e-05,
+      "loss": 1.2623,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1674232152331545,
+      "grad_norm": 0.2643260061740875,
+      "learning_rate": 4.915665314538688e-05,
+      "loss": 1.2092,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16758042013008706,
+      "grad_norm": 0.3190208673477173,
+      "learning_rate": 4.91550611876041e-05,
+      "loss": 1.0543,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16773762502701958,
+      "grad_norm": 0.3194860816001892,
+      "learning_rate": 4.9153467754522095e-05,
+      "loss": 1.1393,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16789482992395213,
+      "grad_norm": 0.2031661570072174,
+      "learning_rate": 4.915187284623817e-05,
+      "loss": 1.2136,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16805203482088468,
+      "grad_norm": 0.2047189325094223,
+      "learning_rate": 4.915027646284974e-05,
+      "loss": 1.2962,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1682092397178172,
+      "grad_norm": 0.31565096974372864,
+      "learning_rate": 4.9148678604454325e-05,
+      "loss": 1.1979,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16836644461474976,
+      "grad_norm": 0.23875312507152557,
+      "learning_rate": 4.914707927114949e-05,
+      "loss": 1.3002,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16852364951168228,
+      "grad_norm": 0.24424909055233002,
+      "learning_rate": 4.9145478463032924e-05,
+      "loss": 1.1491,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16868085440861483,
+      "grad_norm": 0.28339752554893494,
+      "learning_rate": 4.91438761802024e-05,
+      "loss": 1.2398,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16883805930554738,
+      "grad_norm": 0.2435888648033142,
+      "learning_rate": 4.9142272422755786e-05,
+      "loss": 1.3292,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1689952642024799,
+      "grad_norm": 0.21540984511375427,
+      "learning_rate": 4.9140667190791026e-05,
+      "loss": 1.3665,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16915246909941245,
+      "grad_norm": 0.2556820809841156,
+      "learning_rate": 4.913906048440617e-05,
+      "loss": 1.2557,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16930967399634497,
+      "grad_norm": 0.23769475519657135,
+      "learning_rate": 4.913745230369934e-05,
+      "loss": 1.2163,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16946687889327752,
+      "grad_norm": 0.31578120589256287,
+      "learning_rate": 4.913584264876875e-05,
+      "loss": 1.3176,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16962408379021007,
+      "grad_norm": 0.22278232872486115,
+      "learning_rate": 4.913423151971273e-05,
+      "loss": 1.2206,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1697812886871426,
+      "grad_norm": 0.31810736656188965,
+      "learning_rate": 4.913261891662967e-05,
+      "loss": 1.2254,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16993849358407515,
+      "grad_norm": 0.22623823583126068,
+      "learning_rate": 4.913100483961807e-05,
+      "loss": 1.208,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1700956984810077,
+      "grad_norm": 0.27108776569366455,
+      "learning_rate": 4.9129389288776504e-05,
+      "loss": 1.2989,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17025290337794022,
+      "grad_norm": 0.194550558924675,
+      "learning_rate": 4.912777226420365e-05,
+      "loss": 1.3849,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17041010827487277,
+      "grad_norm": 0.20856213569641113,
+      "learning_rate": 4.912615376599826e-05,
+      "loss": 1.2736,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1705673131718053,
+      "grad_norm": 0.19355203211307526,
+      "learning_rate": 4.91245337942592e-05,
+      "loss": 1.256,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17072451806873784,
+      "grad_norm": 0.21832303702831268,
+      "learning_rate": 4.9122912349085395e-05,
+      "loss": 1.1987,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1708817229656704,
+      "grad_norm": 0.22642913460731506,
+      "learning_rate": 4.912128943057589e-05,
+      "loss": 1.3043,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1710389278626029,
+      "grad_norm": 0.22713351249694824,
+      "learning_rate": 4.911966503882981e-05,
+      "loss": 1.1951,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17119613275953546,
+      "grad_norm": 0.29707837104797363,
+      "learning_rate": 4.911803917394634e-05,
+      "loss": 1.2674,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17135333765646799,
+      "grad_norm": 0.27017104625701904,
+      "learning_rate": 4.911641183602481e-05,
+      "loss": 1.1727,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17151054255340054,
+      "grad_norm": 0.23867738246917725,
+      "learning_rate": 4.911478302516461e-05,
+      "loss": 1.2061,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17166774745033309,
+      "grad_norm": 0.26270055770874023,
+      "learning_rate": 4.911315274146521e-05,
+      "loss": 1.2735,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1718249523472656,
+      "grad_norm": 0.25167661905288696,
+      "learning_rate": 4.911152098502617e-05,
+      "loss": 1.2643,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17198215724419816,
+      "grad_norm": 0.46500882506370544,
+      "learning_rate": 4.9109887755947185e-05,
+      "loss": 1.1743,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1721393621411307,
+      "grad_norm": 0.2189512848854065,
+      "learning_rate": 4.910825305432798e-05,
+      "loss": 1.1232,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17229656703806323,
+      "grad_norm": 0.21188926696777344,
+      "learning_rate": 4.9106616880268405e-05,
+      "loss": 1.2031,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17245377193499578,
+      "grad_norm": 0.2135314792394638,
+      "learning_rate": 4.910497923386839e-05,
+      "loss": 1.2547,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1726109768319283,
+      "grad_norm": 0.27527931332588196,
+      "learning_rate": 4.910334011522796e-05,
+      "loss": 1.119,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17276818172886085,
+      "grad_norm": 0.23516559600830078,
+      "learning_rate": 4.910169952444722e-05,
+      "loss": 1.3006,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1729253866257934,
+      "grad_norm": 0.23166057467460632,
+      "learning_rate": 4.910005746162637e-05,
+      "loss": 1.281,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17308259152272593,
+      "grad_norm": 0.22281822562217712,
+      "learning_rate": 4.9098413926865714e-05,
+      "loss": 1.1526,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17323979641965848,
+      "grad_norm": 0.32062655687332153,
+      "learning_rate": 4.909676892026563e-05,
+      "loss": 1.2388,
+      "step": 1102
+    },
+    {
+      "epoch": 0.173397001316591,
+      "grad_norm": 0.29868969321250916,
+      "learning_rate": 4.909512244192657e-05,
+      "loss": 1.2303,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17355420621352355,
+      "grad_norm": 0.23143270611763,
+      "learning_rate": 4.90934744919491e-05,
+      "loss": 1.2137,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1737114111104561,
+      "grad_norm": 0.2830474376678467,
+      "learning_rate": 4.909182507043389e-05,
+      "loss": 1.2178,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17386861600738862,
+      "grad_norm": 0.22427986562252045,
+      "learning_rate": 4.909017417748166e-05,
+      "loss": 1.3153,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17402582090432117,
+      "grad_norm": 0.2587423622608185,
+      "learning_rate": 4.908852181319326e-05,
+      "loss": 1.2669,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17418302580125372,
+      "grad_norm": 0.24905993044376373,
+      "learning_rate": 4.9086867977669594e-05,
+      "loss": 1.2549,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17434023069818624,
+      "grad_norm": 0.26877379417419434,
+      "learning_rate": 4.908521267101167e-05,
+      "loss": 1.2694,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1744974355951188,
+      "grad_norm": 0.2501152753829956,
+      "learning_rate": 4.9083555893320596e-05,
+      "loss": 1.2241,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17465464049205132,
+      "grad_norm": 0.27815014123916626,
+      "learning_rate": 4.908189764469757e-05,
+      "loss": 1.2152,
+      "step": 1111
+    },
+    {
+      "epoch": 0.17481184538898387,
+      "grad_norm": 0.32891881465911865,
+      "learning_rate": 4.9080237925243856e-05,
+      "loss": 1.2638,
+      "step": 1112
+    },
+    {
+      "epoch": 0.17496905028591642,
+      "grad_norm": 0.2137015461921692,
+      "learning_rate": 4.9078576735060825e-05,
+      "loss": 1.2041,
+      "step": 1113
+    },
+    {
+      "epoch": 0.17512625518284894,
+      "grad_norm": 0.17862486839294434,
+      "learning_rate": 4.907691407424995e-05,
+      "loss": 1.3349,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1752834600797815,
+      "grad_norm": 0.25791284441947937,
+      "learning_rate": 4.907524994291276e-05,
+      "loss": 1.2337,
+      "step": 1115
+    },
+    {
+      "epoch": 0.175440664976714,
+      "grad_norm": 0.24266491830348969,
+      "learning_rate": 4.90735843411509e-05,
+      "loss": 1.0939,
+      "step": 1116
+    },
+    {
+      "epoch": 0.17559786987364656,
+      "grad_norm": 0.2618250250816345,
+      "learning_rate": 4.9071917269066114e-05,
+      "loss": 1.2855,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1757550747705791,
+      "grad_norm": 0.2477390021085739,
+      "learning_rate": 4.9070248726760206e-05,
+      "loss": 1.1675,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17591227966751163,
+      "grad_norm": 0.29105502367019653,
+      "learning_rate": 4.906857871433508e-05,
+      "loss": 1.183,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17606948456444418,
+      "grad_norm": 0.2923283874988556,
+      "learning_rate": 4.906690723189275e-05,
+      "loss": 1.1386,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17606948456444418,
+      "eval_loss": 1.219694972038269,
+      "eval_runtime": 2300.2931,
+      "eval_samples_per_second": 4.025,
+      "eval_steps_per_second": 2.012,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17622668946137673,
+      "grad_norm": 0.3278633952140808,
+      "learning_rate": 4.906523427953529e-05,
+      "loss": 1.1738,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17638389435830926,
+      "grad_norm": 0.31546783447265625,
+      "learning_rate": 4.906355985736488e-05,
+      "loss": 1.0894,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1765410992552418,
+      "grad_norm": 0.28350481390953064,
+      "learning_rate": 4.906188396548379e-05,
+      "loss": 1.2774,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17669830415217433,
+      "grad_norm": 0.21374982595443726,
+      "learning_rate": 4.9060206603994385e-05,
+      "loss": 1.37,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17685550904910688,
+      "grad_norm": 0.2343566119670868,
+      "learning_rate": 4.9058527772999095e-05,
+      "loss": 1.2065,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17701271394603943,
+      "grad_norm": 0.29571887850761414,
+      "learning_rate": 4.905684747260047e-05,
+      "loss": 1.1967,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17716991884297195,
+      "grad_norm": 0.2689303457736969,
+      "learning_rate": 4.905516570290113e-05,
+      "loss": 1.2337,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1773271237399045,
+      "grad_norm": 0.22743673622608185,
+      "learning_rate": 4.90534824640038e-05,
+      "loss": 1.1673,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17748432863683702,
+      "grad_norm": 0.36731019616127014,
+      "learning_rate": 4.905179775601126e-05,
+      "loss": 1.1397,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17764153353376957,
+      "grad_norm": 0.2571149468421936,
+      "learning_rate": 4.905011157902645e-05,
+      "loss": 1.1166,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17779873843070212,
+      "grad_norm": 0.2615256905555725,
+      "learning_rate": 4.904842393315231e-05,
+      "loss": 1.2095,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17795594332763465,
+      "grad_norm": 0.28919360041618347,
+      "learning_rate": 4.904673481849194e-05,
+      "loss": 1.0976,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1781131482245672,
+      "grad_norm": 0.3858489990234375,
+      "learning_rate": 4.90450442351485e-05,
+      "loss": 1.1934,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17827035312149975,
+      "grad_norm": 0.2448245733976364,
+      "learning_rate": 4.904335218322524e-05,
+      "loss": 1.1604,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17842755801843227,
+      "grad_norm": 0.2626294195652008,
+      "learning_rate": 4.9041658662825514e-05,
+      "loss": 1.1301,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17858476291536482,
+      "grad_norm": 0.3016091287136078,
+      "learning_rate": 4.903996367405275e-05,
+      "loss": 1.2579,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17874196781229734,
+      "grad_norm": 0.28168612718582153,
+      "learning_rate": 4.9038267217010455e-05,
+      "loss": 1.1471,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1788991727092299,
+      "grad_norm": 0.29256439208984375,
+      "learning_rate": 4.903656929180228e-05,
+      "loss": 1.1598,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17905637760616244,
+      "grad_norm": 0.19786624610424042,
+      "learning_rate": 4.9034869898531895e-05,
+      "loss": 1.2115,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17921358250309496,
+      "grad_norm": 0.17216260731220245,
+      "learning_rate": 4.9033169037303106e-05,
+      "loss": 1.2471,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17937078740002751,
+      "grad_norm": 0.22571730613708496,
+      "learning_rate": 4.9031466708219785e-05,
+      "loss": 1.2226,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17952799229696004,
+      "grad_norm": 0.25510528683662415,
+      "learning_rate": 4.9029762911385915e-05,
+      "loss": 1.1428,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1796851971938926,
+      "grad_norm": 0.19014020264148712,
+      "learning_rate": 4.902805764690556e-05,
+      "loss": 1.2268,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17984240209082514,
+      "grad_norm": 0.25155729055404663,
+      "learning_rate": 4.902635091488286e-05,
+      "loss": 1.1943,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17999960698775766,
+      "grad_norm": 0.3109387159347534,
+      "learning_rate": 4.902464271542206e-05,
+      "loss": 1.176,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1801568118846902,
+      "grad_norm": 0.2269504815340042,
+      "learning_rate": 4.9022933048627496e-05,
+      "loss": 1.2166,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18031401678162276,
+      "grad_norm": 0.20270425081253052,
+      "learning_rate": 4.902122191460358e-05,
+      "loss": 1.235,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18047122167855528,
+      "grad_norm": 0.2519841194152832,
+      "learning_rate": 4.901950931345481e-05,
+      "loss": 1.2418,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18062842657548783,
+      "grad_norm": 0.1967516988515854,
+      "learning_rate": 4.901779524528582e-05,
+      "loss": 1.2979,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18078563147242036,
+      "grad_norm": 0.21120384335517883,
+      "learning_rate": 4.901607971020127e-05,
+      "loss": 1.1557,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1809428363693529,
+      "grad_norm": 0.31649792194366455,
+      "learning_rate": 4.9014362708305944e-05,
+      "loss": 1.3237,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18110004126628546,
+      "grad_norm": 0.24945318698883057,
+      "learning_rate": 4.901264423970471e-05,
+      "loss": 1.2099,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18125724616321798,
+      "grad_norm": 0.30652904510498047,
+      "learning_rate": 4.901092430450254e-05,
+      "loss": 1.1918,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18141445106015053,
+      "grad_norm": 0.2480253279209137,
+      "learning_rate": 4.900920290280446e-05,
+      "loss": 1.2675,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18157165595708305,
+      "grad_norm": 0.3034304976463318,
+      "learning_rate": 4.900748003471561e-05,
+      "loss": 1.2012,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1817288608540156,
+      "grad_norm": 0.2113679200410843,
+      "learning_rate": 4.900575570034124e-05,
+      "loss": 1.2824,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18188606575094815,
+      "grad_norm": 0.34726831316947937,
+      "learning_rate": 4.9004029899786627e-05,
+      "loss": 1.1426,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18204327064788067,
+      "grad_norm": 0.20344194769859314,
+      "learning_rate": 4.900230263315722e-05,
+      "loss": 1.2096,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18220047554481322,
+      "grad_norm": 0.28635072708129883,
+      "learning_rate": 4.900057390055847e-05,
+      "loss": 1.166,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18235768044174577,
+      "grad_norm": 0.21670344471931458,
+      "learning_rate": 4.8998843702095995e-05,
+      "loss": 1.2103,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1825148853386783,
+      "grad_norm": 0.31661516427993774,
+      "learning_rate": 4.899711203787545e-05,
+      "loss": 1.2345,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18267209023561085,
+      "grad_norm": 0.30255556106567383,
+      "learning_rate": 4.899537890800261e-05,
+      "loss": 1.2342,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18282929513254337,
+      "grad_norm": 0.23636944591999054,
+      "learning_rate": 4.899364431258332e-05,
+      "loss": 1.1685,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18298650002947592,
+      "grad_norm": 0.27452319860458374,
+      "learning_rate": 4.8991908251723524e-05,
+      "loss": 1.1263,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18314370492640847,
+      "grad_norm": 0.28636041283607483,
+      "learning_rate": 4.899017072552926e-05,
+      "loss": 1.1961,
+      "step": 1165
+    },
+    {
+      "epoch": 0.183300909823341,
+      "grad_norm": 0.29220953583717346,
+      "learning_rate": 4.8988431734106635e-05,
+      "loss": 1.2414,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18345811472027354,
+      "grad_norm": 0.20738068222999573,
+      "learning_rate": 4.898669127756188e-05,
+      "loss": 1.1499,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18361531961720606,
+      "grad_norm": 0.19913551211357117,
+      "learning_rate": 4.898494935600127e-05,
+      "loss": 1.3538,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1837725245141386,
+      "grad_norm": 0.256979763507843,
+      "learning_rate": 4.8983205969531234e-05,
+      "loss": 1.1979,
+      "step": 1169
+    },
+    {
+      "epoch": 0.18392972941107116,
+      "grad_norm": 0.26307129859924316,
+      "learning_rate": 4.898146111825821e-05,
+      "loss": 1.2054,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18408693430800369,
+      "grad_norm": 0.2451772540807724,
+      "learning_rate": 4.897971480228879e-05,
+      "loss": 1.1901,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18424413920493624,
+      "grad_norm": 0.3223975896835327,
+      "learning_rate": 4.897796702172962e-05,
+      "loss": 1.1825,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18440134410186879,
+      "grad_norm": 0.34991317987442017,
+      "learning_rate": 4.897621777668746e-05,
+      "loss": 1.1371,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1845585489988013,
+      "grad_norm": 0.2680002748966217,
+      "learning_rate": 4.897446706726915e-05,
+      "loss": 1.2179,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18471575389573386,
+      "grad_norm": 0.21509090065956116,
+      "learning_rate": 4.897271489358159e-05,
+      "loss": 1.1284,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18487295879266638,
+      "grad_norm": 0.20545831322669983,
+      "learning_rate": 4.8970961255731826e-05,
+      "loss": 1.2188,
+      "step": 1176
+    },
+    {
+      "epoch": 0.18503016368959893,
+      "grad_norm": 0.23479585349559784,
+      "learning_rate": 4.896920615382695e-05,
+      "loss": 1.2947,
+      "step": 1177
+    },
+    {
+      "epoch": 0.18518736858653148,
+      "grad_norm": 0.2880757749080658,
+      "learning_rate": 4.896744958797417e-05,
+      "loss": 1.1443,
+      "step": 1178
+    },
+    {
+      "epoch": 0.185344573483464,
+      "grad_norm": 0.2431318610906601,
+      "learning_rate": 4.8965691558280744e-05,
+      "loss": 1.1123,
+      "step": 1179
+    },
+    {
+      "epoch": 0.18550177838039655,
+      "grad_norm": 0.21252453327178955,
+      "learning_rate": 4.896393206485407e-05,
+      "loss": 1.326,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18565898327732908,
+      "grad_norm": 0.28821709752082825,
+      "learning_rate": 4.8962171107801596e-05,
+      "loss": 1.1508,
+      "step": 1181
+    },
+    {
+      "epoch": 0.18581618817426163,
+      "grad_norm": 0.2636358141899109,
+      "learning_rate": 4.8960408687230886e-05,
+      "loss": 1.1061,
+      "step": 1182
+    },
+    {
+      "epoch": 0.18597339307119418,
+      "grad_norm": 0.23121225833892822,
+      "learning_rate": 4.895864480324957e-05,
+      "loss": 1.2486,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1861305979681267,
+      "grad_norm": 0.29034245014190674,
+      "learning_rate": 4.895687945596539e-05,
+      "loss": 1.186,
+      "step": 1184
+    },
+    {
+      "epoch": 0.18628780286505925,
+      "grad_norm": 0.3220363259315491,
+      "learning_rate": 4.895511264548617e-05,
+      "loss": 1.1727,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1864450077619918,
+      "grad_norm": 0.2863159477710724,
+      "learning_rate": 4.89533443719198e-05,
+      "loss": 1.1946,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18660221265892432,
+      "grad_norm": 0.27671483159065247,
+      "learning_rate": 4.89515746353743e-05,
+      "loss": 1.2271,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18675941755585687,
+      "grad_norm": 0.2535041570663452,
+      "learning_rate": 4.894980343595775e-05,
+      "loss": 1.2437,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1869166224527894,
+      "grad_norm": 0.34405645728111267,
+      "learning_rate": 4.894803077377833e-05,
+      "loss": 1.1397,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18707382734972194,
+      "grad_norm": 0.28299692273139954,
+      "learning_rate": 4.8946256648944307e-05,
+      "loss": 1.1215,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1872310322466545,
+      "grad_norm": 0.1962118297815323,
+      "learning_rate": 4.8944481061564035e-05,
+      "loss": 1.1908,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18738823714358702,
+      "grad_norm": 0.24563154578208923,
+      "learning_rate": 4.894270401174597e-05,
+      "loss": 1.2265,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18754544204051957,
+      "grad_norm": 0.22452424466609955,
+      "learning_rate": 4.894092549959862e-05,
+      "loss": 1.1673,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1877026469374521,
+      "grad_norm": 0.1847248673439026,
+      "learning_rate": 4.8939145525230646e-05,
+      "loss": 1.2706,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18785985183438464,
+      "grad_norm": 0.2578265964984894,
+      "learning_rate": 4.893736408875075e-05,
+      "loss": 1.2011,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1880170567313172,
+      "grad_norm": 0.2686786353588104,
+      "learning_rate": 4.893558119026772e-05,
+      "loss": 1.3191,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1881742616282497,
+      "grad_norm": 0.27492383122444153,
+      "learning_rate": 4.893379682989047e-05,
+      "loss": 1.1755,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18833146652518226,
+      "grad_norm": 0.2544412612915039,
+      "learning_rate": 4.8932011007727965e-05,
+      "loss": 1.1842,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1884886714221148,
+      "grad_norm": 0.24790935218334198,
+      "learning_rate": 4.893022372388928e-05,
+      "loss": 1.2408,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18864587631904733,
+      "grad_norm": 0.2788006067276001,
+      "learning_rate": 4.892843497848358e-05,
+      "loss": 1.2671,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18880308121597988,
+      "grad_norm": 0.2571476101875305,
+      "learning_rate": 4.892664477162012e-05,
+      "loss": 1.1894,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1889602861129124,
+      "grad_norm": 0.22788426280021667,
+      "learning_rate": 4.892485310340822e-05,
+      "loss": 1.2261,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18911749100984496,
+      "grad_norm": 0.2010507732629776,
+      "learning_rate": 4.892305997395733e-05,
+      "loss": 1.2399,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1892746959067775,
+      "grad_norm": 0.23946425318717957,
+      "learning_rate": 4.892126538337696e-05,
+      "loss": 1.2727,
+      "step": 1204
+    },
+    {
+      "epoch": 0.18943190080371003,
+      "grad_norm": 0.2885929346084595,
+      "learning_rate": 4.8919469331776714e-05,
+      "loss": 1.2376,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18958910570064258,
+      "grad_norm": 0.31879860162734985,
+      "learning_rate": 4.891767181926629e-05,
+      "loss": 1.22,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1897463105975751,
+      "grad_norm": 0.2895459532737732,
+      "learning_rate": 4.891587284595546e-05,
+      "loss": 1.2387,
+      "step": 1207
+    },
+    {
+      "epoch": 0.18990351549450765,
+      "grad_norm": 0.27507272362709045,
+      "learning_rate": 4.891407241195412e-05,
+      "loss": 1.1723,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1900607203914402,
+      "grad_norm": 0.26780039072036743,
+      "learning_rate": 4.8912270517372224e-05,
+      "loss": 1.1549,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19021792528837272,
+      "grad_norm": 0.1915176510810852,
+      "learning_rate": 4.8910467162319826e-05,
+      "loss": 1.109,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19037513018530527,
+      "grad_norm": 0.25054261088371277,
+      "learning_rate": 4.8908662346907064e-05,
+      "loss": 1.1197,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19053233508223782,
+      "grad_norm": 0.24239963293075562,
+      "learning_rate": 4.8906856071244176e-05,
+      "loss": 1.2614,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19068953997917035,
+      "grad_norm": 0.21543578803539276,
+      "learning_rate": 4.890504833544147e-05,
+      "loss": 1.3804,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1908467448761029,
+      "grad_norm": 0.2045610100030899,
+      "learning_rate": 4.8903239139609376e-05,
+      "loss": 1.2108,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19100394977303542,
+      "grad_norm": 0.2209930568933487,
+      "learning_rate": 4.890142848385838e-05,
+      "loss": 1.2329,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19116115466996797,
+      "grad_norm": 0.24921675026416779,
+      "learning_rate": 4.889961636829906e-05,
+      "loss": 1.2009,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19131835956690052,
+      "grad_norm": 0.2356979250907898,
+      "learning_rate": 4.8897802793042115e-05,
+      "loss": 1.211,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19147556446383304,
+      "grad_norm": 0.20199252665042877,
+      "learning_rate": 4.88959877581983e-05,
+      "loss": 1.18,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1916327693607656,
+      "grad_norm": 0.24907195568084717,
+      "learning_rate": 4.889417126387846e-05,
+      "loss": 1.2438,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19178997425769811,
+      "grad_norm": 0.2976427674293518,
+      "learning_rate": 4.889235331019356e-05,
+      "loss": 1.1526,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19194717915463066,
+      "grad_norm": 0.25074872374534607,
+      "learning_rate": 4.889053389725463e-05,
+      "loss": 1.1805,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19210438405156322,
+      "grad_norm": 0.2157672792673111,
+      "learning_rate": 4.8888713025172776e-05,
+      "loss": 1.2103,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19226158894849574,
+      "grad_norm": 0.24573171138763428,
+      "learning_rate": 4.888689069405923e-05,
+      "loss": 1.1981,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1924187938454283,
+      "grad_norm": 0.294160932302475,
+      "learning_rate": 4.888506690402528e-05,
+      "loss": 1.2667,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19257599874236084,
+      "grad_norm": 0.8444136381149292,
+      "learning_rate": 4.8883241655182314e-05,
+      "loss": 1.1977,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19273320363929336,
+      "grad_norm": 0.4191160798072815,
+      "learning_rate": 4.888141494764182e-05,
+      "loss": 1.1981,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1928904085362259,
+      "grad_norm": 0.31621554493904114,
+      "learning_rate": 4.8879586781515376e-05,
+      "loss": 1.2224,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19304761343315843,
+      "grad_norm": 0.2715776860713959,
+      "learning_rate": 4.887775715691462e-05,
+      "loss": 1.1029,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19320481833009098,
+      "grad_norm": 0.2641848623752594,
+      "learning_rate": 4.88759260739513e-05,
+      "loss": 1.1738,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19336202322702353,
+      "grad_norm": 0.2537270188331604,
+      "learning_rate": 4.887409353273727e-05,
+      "loss": 1.2847,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19351922812395606,
+      "grad_norm": 0.2998782694339752,
+      "learning_rate": 4.8872259533384423e-05,
+      "loss": 1.1814,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1936764330208886,
+      "grad_norm": 0.2254815697669983,
+      "learning_rate": 4.8870424076004806e-05,
+      "loss": 1.2004,
+      "step": 1232
+    },
+    {
+      "epoch": 0.19383363791782113,
+      "grad_norm": 0.3711993396282196,
+      "learning_rate": 4.88685871607105e-05,
+      "loss": 1.1502,
+      "step": 1233
+    },
+    {
+      "epoch": 0.19399084281475368,
+      "grad_norm": 0.24783778190612793,
+      "learning_rate": 4.886674878761371e-05,
+      "loss": 1.1185,
+      "step": 1234
+    },
+    {
+      "epoch": 0.19414804771168623,
+      "grad_norm": 0.1896362453699112,
+      "learning_rate": 4.88649089568267e-05,
+      "loss": 1.1856,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19430525260861875,
+      "grad_norm": 0.28106558322906494,
+      "learning_rate": 4.886306766846187e-05,
+      "loss": 1.2196,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1944624575055513,
+      "grad_norm": 0.3023208975791931,
+      "learning_rate": 4.8861224922631645e-05,
+      "loss": 1.1836,
+      "step": 1237
+    },
+    {
+      "epoch": 0.19461966240248385,
+      "grad_norm": 0.36752450466156006,
+      "learning_rate": 4.8859380719448596e-05,
+      "loss": 1.1831,
+      "step": 1238
+    },
+    {
+      "epoch": 0.19477686729941637,
+      "grad_norm": 0.2593975365161896,
+      "learning_rate": 4.885753505902535e-05,
+      "loss": 1.1955,
+      "step": 1239
+    },
+    {
+      "epoch": 0.19493407219634892,
+      "grad_norm": 0.2952882647514343,
+      "learning_rate": 4.885568794147463e-05,
+      "loss": 1.108,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19509127709328145,
+      "grad_norm": 0.2335767149925232,
+      "learning_rate": 4.885383936690926e-05,
+      "loss": 1.2389,
+      "step": 1241
+    },
+    {
+      "epoch": 0.195248481990214,
+      "grad_norm": 0.3618619441986084,
+      "learning_rate": 4.885198933544214e-05,
+      "loss": 1.0247,
+      "step": 1242
+    },
+    {
+      "epoch": 0.19540568688714655,
+      "grad_norm": 0.26691627502441406,
+      "learning_rate": 4.885013784718626e-05,
+      "loss": 1.1516,
+      "step": 1243
+    },
+    {
+      "epoch": 0.19556289178407907,
+      "grad_norm": 0.2977723777294159,
+      "learning_rate": 4.8848284902254705e-05,
+      "loss": 1.1617,
+      "step": 1244
+    },
+    {
+      "epoch": 0.19572009668101162,
+      "grad_norm": 0.33515632152557373,
+      "learning_rate": 4.884643050076064e-05,
+      "loss": 1.1789,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19587730157794414,
+      "grad_norm": 0.275840163230896,
+      "learning_rate": 4.8844574642817334e-05,
+      "loss": 1.1103,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1960345064748767,
+      "grad_norm": 0.26756566762924194,
+      "learning_rate": 4.884271732853813e-05,
+      "loss": 1.2101,
+      "step": 1247
+    },
+    {
+      "epoch": 0.19619171137180924,
+      "grad_norm": 0.20770548284053802,
+      "learning_rate": 4.884085855803647e-05,
+      "loss": 1.2506,
+      "step": 1248
+    },
+    {
+      "epoch": 0.19634891626874176,
+      "grad_norm": 0.2700664699077606,
+      "learning_rate": 4.883899833142588e-05,
+      "loss": 1.2034,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1965061211656743,
+      "grad_norm": 0.2403496950864792,
+      "learning_rate": 4.883713664881997e-05,
+      "loss": 1.1622,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19666332606260686,
+      "grad_norm": 0.2710270881652832,
+      "learning_rate": 4.883527351033245e-05,
+      "loss": 1.0679,
+      "step": 1251
+    },
+    {
+      "epoch": 0.19682053095953939,
+      "grad_norm": 0.2600773870944977,
+      "learning_rate": 4.8833408916077104e-05,
+      "loss": 1.3343,
+      "step": 1252
+    },
+    {
+      "epoch": 0.19697773585647194,
+      "grad_norm": 0.25740665197372437,
+      "learning_rate": 4.883154286616783e-05,
+      "loss": 1.2206,
+      "step": 1253
+    },
+    {
+      "epoch": 0.19713494075340446,
+      "grad_norm": 0.3393601179122925,
+      "learning_rate": 4.8829675360718585e-05,
+      "loss": 1.1518,
+      "step": 1254
+    },
+    {
+      "epoch": 0.197292145650337,
+      "grad_norm": 0.2968616783618927,
+      "learning_rate": 4.8827806399843444e-05,
+      "loss": 1.2547,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19744935054726956,
+      "grad_norm": 0.24990178644657135,
+      "learning_rate": 4.8825935983656535e-05,
+      "loss": 1.2733,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19760655544420208,
+      "grad_norm": 0.31955957412719727,
+      "learning_rate": 4.882406411227212e-05,
+      "loss": 1.2138,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19776376034113463,
+      "grad_norm": 0.22445374727249146,
+      "learning_rate": 4.88221907858045e-05,
+      "loss": 1.1845,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19792096523806715,
+      "grad_norm": 0.32888510823249817,
+      "learning_rate": 4.8820316004368116e-05,
+      "loss": 1.2339,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1980781701349997,
+      "grad_norm": 0.29760921001434326,
+      "learning_rate": 4.8818439768077456e-05,
+      "loss": 1.2216,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19823537503193225,
+      "grad_norm": 0.19965974986553192,
+      "learning_rate": 4.881656207704712e-05,
+      "loss": 1.2608,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19839257992886478,
+      "grad_norm": 0.2538587749004364,
+      "learning_rate": 4.881468293139179e-05,
+      "loss": 1.1989,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19854978482579733,
+      "grad_norm": 0.35299167037010193,
+      "learning_rate": 4.8812802331226224e-05,
+      "loss": 1.1426,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19870698972272988,
+      "grad_norm": 0.3230816423892975,
+      "learning_rate": 4.8810920276665306e-05,
+      "loss": 1.2546,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1988641946196624,
+      "grad_norm": 0.3077559769153595,
+      "learning_rate": 4.880903676782397e-05,
+      "loss": 1.1661,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19902139951659495,
+      "grad_norm": 0.32157936692237854,
+      "learning_rate": 4.8807151804817254e-05,
+      "loss": 1.2141,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19917860441352747,
+      "grad_norm": 0.32653504610061646,
+      "learning_rate": 4.880526538776029e-05,
+      "loss": 1.0623,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19933580931046002,
+      "grad_norm": 0.2675210237503052,
+      "learning_rate": 4.880337751676828e-05,
+      "loss": 1.1408,
+      "step": 1268
+    },
+    {
+      "epoch": 0.19949301420739257,
+      "grad_norm": 0.28380653262138367,
+      "learning_rate": 4.880148819195654e-05,
+      "loss": 1.223,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1996502191043251,
+      "grad_norm": 0.2532847821712494,
+      "learning_rate": 4.8799597413440466e-05,
+      "loss": 1.2133,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19980742400125764,
+      "grad_norm": 0.2972438633441925,
+      "learning_rate": 4.8797705181335526e-05,
+      "loss": 1.2806,
+      "step": 1271
+    },
+    {
+      "epoch": 0.19996462889819017,
+      "grad_norm": 0.2725450098514557,
+      "learning_rate": 4.8795811495757306e-05,
+      "loss": 1.1627,
+      "step": 1272
+    },
+    {
+      "epoch": 0.20012183379512272,
+      "grad_norm": 0.2451506108045578,
+      "learning_rate": 4.879391635682145e-05,
+      "loss": 1.3242,
+      "step": 1273
+    },
+    {
+      "epoch": 0.20027903869205527,
+      "grad_norm": 0.22880415618419647,
+      "learning_rate": 4.8792019764643714e-05,
+      "loss": 1.1535,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2004362435889878,
+      "grad_norm": 0.22470681369304657,
+      "learning_rate": 4.8790121719339935e-05,
+      "loss": 1.268,
+      "step": 1275
+    },
+    {
+      "epoch": 0.20059344848592034,
+      "grad_norm": 0.2413133829832077,
+      "learning_rate": 4.878822222102604e-05,
+      "loss": 1.2291,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2007506533828529,
+      "grad_norm": 0.23373375833034515,
+      "learning_rate": 4.878632126981804e-05,
+      "loss": 1.1007,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2009078582797854,
+      "grad_norm": 0.3018023371696472,
+      "learning_rate": 4.878441886583203e-05,
+      "loss": 1.2393,
+      "step": 1278
+    },
+    {
+      "epoch": 0.20106506317671796,
+      "grad_norm": 0.2107972353696823,
+      "learning_rate": 4.878251500918421e-05,
+      "loss": 1.3164,
+      "step": 1279
+    },
+    {
+      "epoch": 0.20122226807365048,
+      "grad_norm": 0.24787524342536926,
+      "learning_rate": 4.878060969999087e-05,
+      "loss": 1.217,
+      "step": 1280
+    },
+    {
+      "epoch": 0.20122226807365048,
+      "eval_loss": 1.2021143436431885,
+      "eval_runtime": 2276.1827,
+      "eval_samples_per_second": 4.067,
+      "eval_steps_per_second": 2.034,
+      "step": 1280
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 12722,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 160,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.55664759605035e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}