diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5641 @@
+{
+  "best_metric": 2.631765127182007,
+  "best_model_checkpoint": "Meta-Llama-3.1-8B-Instruct-finetuned/checkpoint-800",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00125,
+      "grad_norm": 0.8839266300201416,
+      "learning_rate": 1.9975e-05,
+      "loss": 3.5886,
+      "step": 1
+    },
+    {
+      "epoch": 0.0025,
+      "grad_norm": 0.5743476748466492,
+      "learning_rate": 1.9950000000000004e-05,
+      "loss": 3.1499,
+      "step": 2
+    },
+    {
+      "epoch": 0.00375,
+      "grad_norm": 0.6939172148704529,
+      "learning_rate": 1.9925e-05,
+      "loss": 2.4424,
+      "step": 3
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 0.35782289505004883,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 1.6857,
+      "step": 4
+    },
+    {
+      "epoch": 0.00625,
+      "grad_norm": 0.30019086599349976,
+      "learning_rate": 1.9875000000000002e-05,
+      "loss": 2.4306,
+      "step": 5
+    },
+    {
+      "epoch": 0.0075,
+      "grad_norm": 0.7822341918945312,
+      "learning_rate": 1.985e-05,
+      "loss": 3.1167,
+      "step": 6
+    },
+    {
+      "epoch": 0.00875,
+      "grad_norm": 1.0825086832046509,
+      "learning_rate": 1.9825e-05,
+      "loss": 3.4788,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6958524584770203,
+      "learning_rate": 1.98e-05,
+      "loss": 2.4278,
+      "step": 8
+    },
+    {
+      "epoch": 0.01125,
+      "grad_norm": 1.0385024547576904,
+      "learning_rate": 1.9775000000000003e-05,
+      "loss": 2.7115,
+      "step": 9
+    },
+    {
+      "epoch": 0.0125,
+      "grad_norm": 1.0596249103546143,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 3.8705,
+      "step": 10
+    },
+    {
+      "epoch": 0.01375,
+      "grad_norm": 1.014846682548523,
+      "learning_rate": 1.9725000000000002e-05,
+      "loss": 2.7525,
+      "step": 11
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 1.0246942043304443,
+      "learning_rate": 1.97e-05,
+      "loss": 3.2441,
+      "step": 12
+    },
+    {
+      "epoch": 0.01625,
+      "grad_norm": 0.5323541760444641,
+      "learning_rate": 1.9675e-05,
+      "loss": 2.2139,
+      "step": 13
+    },
+    {
+      "epoch": 0.0175,
+      "grad_norm": 0.6004053950309753,
+      "learning_rate": 1.9650000000000003e-05,
+      "loss": 2.7995,
+      "step": 14
+    },
+    {
+      "epoch": 0.01875,
+      "grad_norm": 0.6828798651695251,
+      "learning_rate": 1.9625e-05,
+      "loss": 2.7872,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8382136225700378,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.9668,
+      "step": 16
+    },
+    {
+      "epoch": 0.02125,
+      "grad_norm": 0.5875881910324097,
+      "learning_rate": 1.9575e-05,
+      "loss": 2.4291,
+      "step": 17
+    },
+    {
+      "epoch": 0.0225,
+      "grad_norm": 1.260549545288086,
+      "learning_rate": 1.955e-05,
+      "loss": 3.2646,
+      "step": 18
+    },
+    {
+      "epoch": 0.02375,
+      "grad_norm": 0.595811128616333,
+      "learning_rate": 1.9525e-05,
+      "loss": 2.5531,
+      "step": 19
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 0.5156136155128479,
+      "learning_rate": 1.95e-05,
+      "loss": 2.4804,
+      "step": 20
+    },
+    {
+      "epoch": 0.02625,
+      "grad_norm": 0.4905349314212799,
+      "learning_rate": 1.9475000000000002e-05,
+      "loss": 2.1618,
+      "step": 21
+    },
+    {
+      "epoch": 0.0275,
+      "grad_norm": 0.3786165118217468,
+      "learning_rate": 1.9450000000000002e-05,
+      "loss": 2.2184,
+      "step": 22
+    },
+    {
+      "epoch": 0.02875,
+      "grad_norm": 0.6827333569526672,
+      "learning_rate": 1.9425e-05,
+      "loss": 2.2265,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9549139738082886,
+      "learning_rate": 1.94e-05,
+      "loss": 3.0346,
+      "step": 24
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 0.4533088207244873,
+      "learning_rate": 1.9375e-05,
+      "loss": 1.9096,
+      "step": 25
+    },
+    {
+      "epoch": 0.0325,
+      "grad_norm": 1.1265308856964111,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 3.1981,
+      "step": 26
+    },
+    {
+      "epoch": 0.03375,
+      "grad_norm": 0.6239346265792847,
+      "learning_rate": 1.9325000000000002e-05,
+      "loss": 2.6925,
+      "step": 27
+    },
+    {
+      "epoch": 0.035,
+      "grad_norm": 0.8297908306121826,
+      "learning_rate": 1.93e-05,
+      "loss": 2.9377,
+      "step": 28
+    },
+    {
+      "epoch": 0.03625,
+      "grad_norm": 0.5328755974769592,
+      "learning_rate": 1.9275e-05,
+      "loss": 3.5033,
+      "step": 29
+    },
+    {
+      "epoch": 0.0375,
+      "grad_norm": 0.748451292514801,
+      "learning_rate": 1.925e-05,
+      "loss": 3.3768,
+      "step": 30
+    },
+    {
+      "epoch": 0.03875,
+      "grad_norm": 0.6435258388519287,
+      "learning_rate": 1.9225000000000003e-05,
+      "loss": 2.8508,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5679820775985718,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 2.6533,
+      "step": 32
+    },
+    {
+      "epoch": 0.04125,
+      "grad_norm": 0.8853691220283508,
+      "learning_rate": 1.9175000000000002e-05,
+      "loss": 3.0708,
+      "step": 33
+    },
+    {
+      "epoch": 0.0425,
+      "grad_norm": 0.6417959332466125,
+      "learning_rate": 1.915e-05,
+      "loss": 1.6389,
+      "step": 34
+    },
+    {
+      "epoch": 0.04375,
+      "grad_norm": 0.5146605372428894,
+      "learning_rate": 1.9125000000000004e-05,
+      "loss": 2.7314,
+      "step": 35
+    },
+    {
+      "epoch": 0.045,
+      "grad_norm": 1.4297131299972534,
+      "learning_rate": 1.91e-05,
+      "loss": 3.6263,
+      "step": 36
+    },
+    {
+      "epoch": 0.04625,
+      "grad_norm": 1.8327345848083496,
+      "learning_rate": 1.9075000000000003e-05,
+      "loss": 3.2042,
+      "step": 37
+    },
+    {
+      "epoch": 0.0475,
+      "grad_norm": 0.9266634583473206,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 2.8001,
+      "step": 38
+    },
+    {
+      "epoch": 0.04875,
+      "grad_norm": 0.944769024848938,
+      "learning_rate": 1.9025e-05,
+      "loss": 2.8379,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6269248723983765,
+      "learning_rate": 1.9e-05,
+      "loss": 2.5707,
+      "step": 40
+    },
+    {
+      "epoch": 0.05125,
+      "grad_norm": 0.8288752436637878,
+      "learning_rate": 1.8975e-05,
+      "loss": 3.1964,
+      "step": 41
+    },
+    {
+      "epoch": 0.0525,
+      "grad_norm": 1.4263437986373901,
+      "learning_rate": 1.8950000000000003e-05,
+      "loss": 3.5353,
+      "step": 42
+    },
+    {
+      "epoch": 0.05375,
+      "grad_norm": 1.2455209493637085,
+      "learning_rate": 1.8925000000000003e-05,
+      "loss": 2.7257,
+      "step": 43
+    },
+    {
+      "epoch": 0.055,
+      "grad_norm": 0.5519097447395325,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 2.2432,
+      "step": 44
+    },
+    {
+      "epoch": 0.05625,
+      "grad_norm": 0.7771982550621033,
+      "learning_rate": 1.8875e-05,
+      "loss": 2.1759,
+      "step": 45
+    },
+    {
+      "epoch": 0.0575,
+      "grad_norm": 1.3399147987365723,
+      "learning_rate": 1.885e-05,
+      "loss": 3.1654,
+      "step": 46
+    },
+    {
+      "epoch": 0.05875,
+      "grad_norm": 0.4468587040901184,
+      "learning_rate": 1.8825000000000004e-05,
+      "loss": 2.0097,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6694964170455933,
+      "learning_rate": 1.88e-05,
+      "loss": 2.3989,
+      "step": 48
+    },
+    {
+      "epoch": 0.06125,
+      "grad_norm": 0.583661675453186,
+      "learning_rate": 1.8775000000000002e-05,
+      "loss": 2.3641,
+      "step": 49
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 1.7708433866500854,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 3.4434,
+      "step": 50
+    },
+    {
+      "epoch": 0.06375,
+      "grad_norm": 1.0765807628631592,
+      "learning_rate": 1.8725e-05,
+      "loss": 3.0016,
+      "step": 51
+    },
+    {
+      "epoch": 0.065,
+      "grad_norm": 2.168282985687256,
+      "learning_rate": 1.8700000000000004e-05,
+      "loss": 3.9576,
+      "step": 52
+    },
+    {
+      "epoch": 0.06625,
+      "grad_norm": 1.4953194856643677,
+      "learning_rate": 1.8675e-05,
+      "loss": 3.1084,
+      "step": 53
+    },
+    {
+      "epoch": 0.0675,
+      "grad_norm": 0.9680989980697632,
+      "learning_rate": 1.8650000000000003e-05,
+      "loss": 2.678,
+      "step": 54
+    },
+    {
+      "epoch": 0.06875,
+      "grad_norm": 0.6683467030525208,
+      "learning_rate": 1.8625000000000002e-05,
+      "loss": 2.5912,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5850144624710083,
+      "learning_rate": 1.86e-05,
+      "loss": 2.5303,
+      "step": 56
+    },
+    {
+      "epoch": 0.07125,
+      "grad_norm": 1.957913875579834,
+      "learning_rate": 1.8575e-05,
+      "loss": 3.5738,
+      "step": 57
+    },
+    {
+      "epoch": 0.0725,
+      "grad_norm": 0.4880407154560089,
+      "learning_rate": 1.855e-05,
+      "loss": 2.4431,
+      "step": 58
+    },
+    {
+      "epoch": 0.07375,
+      "grad_norm": 0.6582357883453369,
+      "learning_rate": 1.8525000000000003e-05,
+      "loss": 2.7631,
+      "step": 59
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 1.223876953125,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 2.551,
+      "step": 60
+    },
+    {
+      "epoch": 0.07625,
+      "grad_norm": 0.7573391199111938,
+      "learning_rate": 1.8475000000000002e-05,
+      "loss": 2.7722,
+      "step": 61
+    },
+    {
+      "epoch": 0.0775,
+      "grad_norm": 1.3893153667449951,
+      "learning_rate": 1.845e-05,
+      "loss": 2.6213,
+      "step": 62
+    },
+    {
+      "epoch": 0.07875,
+      "grad_norm": 0.8527727723121643,
+      "learning_rate": 1.8425e-05,
+      "loss": 2.76,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.539513885974884,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 2.4052,
+      "step": 64
+    },
+    {
+      "epoch": 0.08125,
+      "grad_norm": 1.2717337608337402,
+      "learning_rate": 1.8375e-05,
+      "loss": 3.2137,
+      "step": 65
+    },
+    {
+      "epoch": 0.0825,
+      "grad_norm": 0.9680526852607727,
+      "learning_rate": 1.8350000000000002e-05,
+      "loss": 1.867,
+      "step": 66
+    },
+    {
+      "epoch": 0.08375,
+      "grad_norm": 0.7457209825515747,
+      "learning_rate": 1.8325e-05,
+      "loss": 2.9906,
+      "step": 67
+    },
+    {
+      "epoch": 0.085,
+      "grad_norm": 0.9684948325157166,
+      "learning_rate": 1.83e-05,
+      "loss": 2.3277,
+      "step": 68
+    },
+    {
+      "epoch": 0.08625,
+      "grad_norm": 1.1279115676879883,
+      "learning_rate": 1.8275e-05,
+      "loss": 2.9131,
+      "step": 69
+    },
+    {
+      "epoch": 0.0875,
+      "grad_norm": 0.7532200217247009,
+      "learning_rate": 1.825e-05,
+      "loss": 2.848,
+      "step": 70
+    },
+    {
+      "epoch": 0.08875,
+      "grad_norm": 2.1061675548553467,
+      "learning_rate": 1.8225000000000003e-05,
+      "loss": 3.3145,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2701787948608398,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 2.7297,
+      "step": 72
+    },
+    {
+      "epoch": 0.09125,
+      "grad_norm": 0.8566807508468628,
+      "learning_rate": 1.8175e-05,
+      "loss": 2.5054,
+      "step": 73
+    },
+    {
+      "epoch": 0.0925,
+      "grad_norm": 0.7943547964096069,
+      "learning_rate": 1.815e-05,
+      "loss": 2.9178,
+      "step": 74
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.2525556087493896,
+      "learning_rate": 1.8125e-05,
+      "loss": 3.1584,
+      "step": 75
+    },
+    {
+      "epoch": 0.095,
+      "grad_norm": 0.9087671041488647,
+      "learning_rate": 1.8100000000000003e-05,
+      "loss": 2.5073,
+      "step": 76
+    },
+    {
+      "epoch": 0.09625,
+      "grad_norm": 1.253976583480835,
+      "learning_rate": 1.8075000000000002e-05,
+      "loss": 2.8634,
+      "step": 77
+    },
+    {
+      "epoch": 0.0975,
+      "grad_norm": 1.6665199995040894,
+      "learning_rate": 1.805e-05,
+      "loss": 2.1963,
+      "step": 78
+    },
+    {
+      "epoch": 0.09875,
+      "grad_norm": 1.2400460243225098,
+      "learning_rate": 1.8025e-05,
+      "loss": 2.4813,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0923187732696533,
+      "learning_rate": 1.8e-05,
+      "loss": 3.0353,
+      "step": 80
+    },
+    {
+      "epoch": 0.10125,
+      "grad_norm": 0.6350508332252502,
+      "learning_rate": 1.7975000000000003e-05,
+      "loss": 2.4959,
+      "step": 81
+    },
+    {
+      "epoch": 0.1025,
+      "grad_norm": 1.7657043933868408,
+      "learning_rate": 1.795e-05,
+      "loss": 3.2626,
+      "step": 82
+    },
+    {
+      "epoch": 0.10375,
+      "grad_norm": 2.0525500774383545,
+      "learning_rate": 1.7925000000000002e-05,
+      "loss": 2.5613,
+      "step": 83
+    },
+    {
+      "epoch": 0.105,
+      "grad_norm": 0.8630340099334717,
+      "learning_rate": 1.79e-05,
+      "loss": 2.54,
+      "step": 84
+    },
+    {
+      "epoch": 0.10625,
+      "grad_norm": 2.1113481521606445,
+      "learning_rate": 1.7875e-05,
+      "loss": 3.6185,
+      "step": 85
+    },
+    {
+      "epoch": 0.1075,
+      "grad_norm": 1.0695478916168213,
+      "learning_rate": 1.785e-05,
+      "loss": 2.9277,
+      "step": 86
+    },
+    {
+      "epoch": 0.10875,
+      "grad_norm": 1.6546651124954224,
+      "learning_rate": 1.7825e-05,
+      "loss": 2.7956,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6290825605392456,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 2.6147,
+      "step": 88
+    },
+    {
+      "epoch": 0.11125,
+      "grad_norm": 0.8302910923957825,
+      "learning_rate": 1.7775000000000002e-05,
+      "loss": 2.6113,
+      "step": 89
+    },
+    {
+      "epoch": 0.1125,
+      "grad_norm": 2.2139322757720947,
+      "learning_rate": 1.775e-05,
+      "loss": 2.7841,
+      "step": 90
+    },
+    {
+      "epoch": 0.11375,
+      "grad_norm": 0.5742428302764893,
+      "learning_rate": 1.7725e-05,
+      "loss": 2.6969,
+      "step": 91
+    },
+    {
+      "epoch": 0.115,
+      "grad_norm": 1.6150765419006348,
+      "learning_rate": 1.77e-05,
+      "loss": 3.3552,
+      "step": 92
+    },
+    {
+      "epoch": 0.11625,
+      "grad_norm": 1.053296685218811,
+      "learning_rate": 1.7675000000000003e-05,
+      "loss": 2.5297,
+      "step": 93
+    },
+    {
+      "epoch": 0.1175,
+      "grad_norm": 0.8964847922325134,
+      "learning_rate": 1.7650000000000002e-05,
+      "loss": 2.5295,
+      "step": 94
+    },
+    {
+      "epoch": 0.11875,
+      "grad_norm": 0.9291931390762329,
+      "learning_rate": 1.7625e-05,
+      "loss": 2.4275,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3160396814346313,
+      "learning_rate": 1.76e-05,
+      "loss": 2.727,
+      "step": 96
+    },
+    {
+      "epoch": 0.12125,
+      "grad_norm": 3.190565824508667,
+      "learning_rate": 1.7575000000000004e-05,
+      "loss": 3.1598,
+      "step": 97
+    },
+    {
+      "epoch": 0.1225,
+      "grad_norm": 1.1592965126037598,
+      "learning_rate": 1.755e-05,
+      "loss": 1.8949,
+      "step": 98
+    },
+    {
+      "epoch": 0.12375,
+      "grad_norm": 1.942399263381958,
+      "learning_rate": 1.7525000000000002e-05,
+      "loss": 2.8239,
+      "step": 99
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.8165029883384705,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 2.0449,
+      "step": 100
+    },
+    {
+      "epoch": 0.12625,
+      "grad_norm": 0.9424604773521423,
+      "learning_rate": 1.7475e-05,
+      "loss": 2.994,
+      "step": 101
+    },
+    {
+      "epoch": 0.1275,
+      "grad_norm": 2.3983001708984375,
+      "learning_rate": 1.7450000000000004e-05,
+      "loss": 4.2536,
+      "step": 102
+    },
+    {
+      "epoch": 0.12875,
+      "grad_norm": 1.9262495040893555,
+      "learning_rate": 1.7425e-05,
+      "loss": 3.5003,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5252268314361572,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 2.2976,
+      "step": 104
+    },
+    {
+      "epoch": 0.13125,
+      "grad_norm": 1.0399681329727173,
+      "learning_rate": 1.7375000000000002e-05,
+      "loss": 2.5173,
+      "step": 105
+    },
+    {
+      "epoch": 0.1325,
+      "grad_norm": 0.9736959338188171,
+      "learning_rate": 1.735e-05,
+      "loss": 2.475,
+      "step": 106
+    },
+    {
+      "epoch": 0.13375,
+      "grad_norm": 2.253629446029663,
+      "learning_rate": 1.7325e-05,
+      "loss": 3.4539,
+      "step": 107
+    },
+    {
+      "epoch": 0.135,
+      "grad_norm": 0.9584936499595642,
+      "learning_rate": 1.73e-05,
+      "loss": 2.3618,
+      "step": 108
+    },
+    {
+      "epoch": 0.13625,
+      "grad_norm": 0.7614779472351074,
+      "learning_rate": 1.7275000000000003e-05,
+      "loss": 2.6198,
+      "step": 109
+    },
+    {
+      "epoch": 0.1375,
+      "grad_norm": 1.4478983879089355,
+      "learning_rate": 1.7250000000000003e-05,
+      "loss": 2.4151,
+      "step": 110
+    },
+    {
+      "epoch": 0.13875,
+      "grad_norm": 1.6698650121688843,
+      "learning_rate": 1.7225000000000002e-05,
+      "loss": 2.8188,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1940186023712158,
+      "learning_rate": 1.72e-05,
+      "loss": 2.9268,
+      "step": 112
+    },
+    {
+      "epoch": 0.14125,
+      "grad_norm": 1.3393628597259521,
+      "learning_rate": 1.7175e-05,
+      "loss": 3.246,
+      "step": 113
+    },
+    {
+      "epoch": 0.1425,
+      "grad_norm": 1.6196234226226807,
+      "learning_rate": 1.7150000000000004e-05,
+      "loss": 3.1591,
+      "step": 114
+    },
+    {
+      "epoch": 0.14375,
+      "grad_norm": 2.1916654109954834,
+      "learning_rate": 1.7125e-05,
+      "loss": 2.6534,
+      "step": 115
+    },
+    {
+      "epoch": 0.145,
+      "grad_norm": 1.725541591644287,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 2.5497,
+      "step": 116
+    },
+    {
+      "epoch": 0.14625,
+      "grad_norm": 1.0600862503051758,
+      "learning_rate": 1.7075e-05,
+      "loss": 2.5041,
+      "step": 117
+    },
+    {
+      "epoch": 0.1475,
+      "grad_norm": 0.9649932980537415,
+      "learning_rate": 1.705e-05,
+      "loss": 2.8991,
+      "step": 118
+    },
+    {
+      "epoch": 0.14875,
+      "grad_norm": 1.460185170173645,
+      "learning_rate": 1.7025e-05,
+      "loss": 2.7737,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8451898694038391,
+      "learning_rate": 1.7e-05,
+      "loss": 2.3685,
+      "step": 120
+    },
+    {
+      "epoch": 0.15125,
+      "grad_norm": 1.7984133958816528,
+      "learning_rate": 1.6975000000000003e-05,
+      "loss": 2.599,
+      "step": 121
+    },
+    {
+      "epoch": 0.1525,
+      "grad_norm": 0.9329972267150879,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 2.9886,
+      "step": 122
+    },
+    {
+      "epoch": 0.15375,
+      "grad_norm": 0.725104808807373,
+      "learning_rate": 1.6925e-05,
+      "loss": 2.5776,
+      "step": 123
+    },
+    {
+      "epoch": 0.155,
+      "grad_norm": 0.7359060049057007,
+      "learning_rate": 1.69e-05,
+      "loss": 2.8262,
+      "step": 124
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 1.0401649475097656,
+      "learning_rate": 1.6875e-05,
+      "loss": 2.853,
+      "step": 125
+    },
+    {
+      "epoch": 0.1575,
+      "grad_norm": 1.748463749885559,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 2.4642,
+      "step": 126
+    },
+    {
+      "epoch": 0.15875,
+      "grad_norm": 1.278817057609558,
+      "learning_rate": 1.6825000000000002e-05,
+      "loss": 2.1242,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4376240968704224,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 2.7639,
+      "step": 128
+    },
+    {
+      "epoch": 0.16125,
+      "grad_norm": 1.3094562292099,
+      "learning_rate": 1.6775e-05,
+      "loss": 2.4866,
+      "step": 129
+    },
+    {
+      "epoch": 0.1625,
+      "grad_norm": 0.7014777064323425,
+      "learning_rate": 1.675e-05,
+      "loss": 3.0492,
+      "step": 130
+    },
+    {
+      "epoch": 0.16375,
+      "grad_norm": 1.578884243965149,
+      "learning_rate": 1.6725000000000003e-05,
+      "loss": 3.039,
+      "step": 131
+    },
+    {
+      "epoch": 0.165,
+      "grad_norm": 0.9808546900749207,
+      "learning_rate": 1.67e-05,
+      "loss": 2.7078,
+      "step": 132
+    },
+    {
+      "epoch": 0.16625,
+      "grad_norm": 0.7291256189346313,
+      "learning_rate": 1.6675000000000002e-05,
+      "loss": 2.3461,
+      "step": 133
+    },
+    {
+      "epoch": 0.1675,
+      "grad_norm": 0.8226473927497864,
+      "learning_rate": 1.665e-05,
+      "loss": 2.7881,
+      "step": 134
+    },
+    {
+      "epoch": 0.16875,
+      "grad_norm": 1.732428789138794,
+      "learning_rate": 1.6625e-05,
+      "loss": 2.9649,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4389218091964722,
+      "learning_rate": 1.66e-05,
+      "loss": 2.7957,
+      "step": 136
+    },
+    {
+      "epoch": 0.17125,
+      "grad_norm": 1.5826399326324463,
+      "learning_rate": 1.6575e-05,
+      "loss": 2.911,
+      "step": 137
+    },
+    {
+      "epoch": 0.1725,
+      "grad_norm": 1.0410722494125366,
+      "learning_rate": 1.6550000000000002e-05,
+      "loss": 2.6444,
+      "step": 138
+    },
+    {
+      "epoch": 0.17375,
+      "grad_norm": 0.9773198366165161,
+      "learning_rate": 1.6525000000000002e-05,
+      "loss": 2.3028,
+      "step": 139
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 0.9613956212997437,
+      "learning_rate": 1.65e-05,
+      "loss": 2.2155,
+      "step": 140
+    },
+    {
+      "epoch": 0.17625,
+      "grad_norm": 1.3112648725509644,
+      "learning_rate": 1.6475e-05,
+      "loss": 2.2479,
+      "step": 141
+    },
+    {
+      "epoch": 0.1775,
+      "grad_norm": 1.0890480279922485,
+      "learning_rate": 1.645e-05,
+      "loss": 2.155,
+      "step": 142
+    },
+    {
+      "epoch": 0.17875,
+      "grad_norm": 0.837094247341156,
+      "learning_rate": 1.6425000000000003e-05,
+      "loss": 2.6826,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2103476524353027,
+      "learning_rate": 1.64e-05,
+      "loss": 3.228,
+      "step": 144
+    },
+    {
+      "epoch": 0.18125,
+      "grad_norm": 1.9761539697647095,
+      "learning_rate": 1.6375e-05,
+      "loss": 2.7196,
+      "step": 145
+    },
+    {
+      "epoch": 0.1825,
+      "grad_norm": 0.9773359894752502,
+      "learning_rate": 1.635e-05,
+      "loss": 2.9387,
+      "step": 146
+    },
+    {
+      "epoch": 0.18375,
+      "grad_norm": 0.9832570552825928,
+      "learning_rate": 1.6325e-05,
+      "loss": 2.8681,
+      "step": 147
+    },
+    {
+      "epoch": 0.185,
+      "grad_norm": 1.0604499578475952,
+      "learning_rate": 1.63e-05,
+      "loss": 2.5064,
+      "step": 148
+    },
+    {
+      "epoch": 0.18625,
+      "grad_norm": 0.48051193356513977,
+      "learning_rate": 1.6275e-05,
+      "loss": 2.4564,
+      "step": 149
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 1.50802481174469,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 2.8379,
+      "step": 150
+    },
+    {
+      "epoch": 0.18875,
+      "grad_norm": 1.812436580657959,
+      "learning_rate": 1.6225e-05,
+      "loss": 1.9878,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7669923305511475,
+      "learning_rate": 1.62e-05,
+      "loss": 2.8151,
+      "step": 152
+    },
+    {
+      "epoch": 0.19125,
+      "grad_norm": 2.281033754348755,
+      "learning_rate": 1.6175e-05,
+      "loss": 3.3304,
+      "step": 153
+    },
+    {
+      "epoch": 0.1925,
+      "grad_norm": 0.9153265357017517,
+      "learning_rate": 1.6150000000000003e-05,
+      "loss": 1.8299,
+      "step": 154
+    },
+    {
+      "epoch": 0.19375,
+      "grad_norm": 1.041798710823059,
+      "learning_rate": 1.6125000000000002e-05,
+      "loss": 2.9185,
+      "step": 155
+    },
+    {
+      "epoch": 0.195,
+      "grad_norm": 0.6834219694137573,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 2.3889,
+      "step": 156
+    },
+    {
+      "epoch": 0.19625,
+      "grad_norm": 1.8793840408325195,
+      "learning_rate": 1.6075e-05,
+      "loss": 2.7536,
+      "step": 157
+    },
+    {
+      "epoch": 0.1975,
+      "grad_norm": 1.4421665668487549,
+      "learning_rate": 1.605e-05,
+      "loss": 2.9225,
+      "step": 158
+    },
+    {
+      "epoch": 0.19875,
+      "grad_norm": 2.3824641704559326,
+      "learning_rate": 1.6025000000000003e-05,
+      "loss": 3.4679,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.144702672958374,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 3.194,
+      "step": 160
+    },
+    {
+      "epoch": 0.20125,
+      "grad_norm": 1.5149937868118286,
+      "learning_rate": 1.5975000000000002e-05,
+      "loss": 3.0138,
+      "step": 161
+    },
+    {
+      "epoch": 0.2025,
+      "grad_norm": 2.1063072681427,
+      "learning_rate": 1.595e-05,
+      "loss": 3.0519,
+      "step": 162
+    },
+    {
+      "epoch": 0.20375,
+      "grad_norm": 0.820692777633667,
+      "learning_rate": 1.5925e-05,
+      "loss": 1.9736,
+      "step": 163
+    },
+    {
+      "epoch": 0.205,
+      "grad_norm": 1.1006802320480347,
+      "learning_rate": 1.5900000000000004e-05,
+      "loss": 2.7983,
+      "step": 164
+    },
+    {
+      "epoch": 0.20625,
+      "grad_norm": 1.038564920425415,
+      "learning_rate": 1.5875e-05,
+      "loss": 2.506,
+      "step": 165
+    },
+    {
+      "epoch": 0.2075,
+      "grad_norm": 1.1868886947631836,
+      "learning_rate": 1.5850000000000002e-05,
+      "loss": 2.2386,
+      "step": 166
+    },
+    {
+      "epoch": 0.20875,
+      "grad_norm": 0.9923886060714722,
+      "learning_rate": 1.5825000000000002e-05,
+      "loss": 2.2732,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5702191591262817,
+      "learning_rate": 1.58e-05,
+      "loss": 3.1675,
+      "step": 168
+    },
+    {
+      "epoch": 0.21125,
+      "grad_norm": 1.3807926177978516,
+      "learning_rate": 1.5775e-05,
+      "loss": 2.1471,
+      "step": 169
+    },
+    {
+      "epoch": 0.2125,
+      "grad_norm": 1.0479799509048462,
+      "learning_rate": 1.575e-05,
+      "loss": 2.1308,
+      "step": 170
+    },
+    {
+      "epoch": 0.21375,
+      "grad_norm": 1.8413770198822021,
+      "learning_rate": 1.5725000000000003e-05,
+      "loss": 2.9807,
+      "step": 171
+    },
+    {
+      "epoch": 0.215,
+      "grad_norm": 1.8779033422470093,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 2.6333,
+      "step": 172
+    },
+    {
+      "epoch": 0.21625,
+      "grad_norm": 0.9915759563446045,
+      "learning_rate": 1.5675e-05,
+      "loss": 2.1059,
+      "step": 173
+    },
+    {
+      "epoch": 0.2175,
+      "grad_norm": 1.3852919340133667,
+      "learning_rate": 1.565e-05,
+      "loss": 2.8429,
+      "step": 174
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.8641107082366943,
+      "learning_rate": 1.5625e-05,
+      "loss": 2.5064,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1318968534469604,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 2.6325,
+      "step": 176
+    },
+    {
+      "epoch": 0.22125,
+      "grad_norm": 1.3027793169021606,
+      "learning_rate": 1.5575000000000002e-05,
+      "loss": 2.0696,
+      "step": 177
+    },
+    {
+      "epoch": 0.2225,
+      "grad_norm": 1.8702739477157593,
+      "learning_rate": 1.5550000000000002e-05,
+      "loss": 2.6865,
+      "step": 178
+    },
+    {
+      "epoch": 0.22375,
+      "grad_norm": 1.9452624320983887,
+      "learning_rate": 1.5525e-05,
+      "loss": 3.6065,
+      "step": 179
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 2.445535182952881,
+      "learning_rate": 1.55e-05,
+      "loss": 3.6079,
+      "step": 180
+    },
+    {
+      "epoch": 0.22625,
+      "grad_norm": 1.1266591548919678,
+      "learning_rate": 1.5475000000000003e-05,
+      "loss": 2.3907,
+      "step": 181
+    },
+    {
+      "epoch": 0.2275,
+      "grad_norm": 1.3965399265289307,
+      "learning_rate": 1.545e-05,
+      "loss": 2.7698,
+      "step": 182
+    },
+    {
+      "epoch": 0.22875,
+      "grad_norm": 1.1321989297866821,
+      "learning_rate": 1.5425000000000002e-05,
+      "loss": 3.7761,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.289818525314331,
+      "learning_rate": 1.54e-05,
+      "loss": 2.7404,
+      "step": 184
+    },
+    {
+      "epoch": 0.23125,
+      "grad_norm": 1.3494073152542114,
+      "learning_rate": 1.5375e-05,
+      "loss": 2.6953,
+      "step": 185
+    },
+    {
+      "epoch": 0.2325,
+      "grad_norm": 0.6751367449760437,
+      "learning_rate": 1.535e-05,
+      "loss": 2.4991,
+      "step": 186
+    },
+    {
+      "epoch": 0.23375,
+      "grad_norm": 1.2142468690872192,
+      "learning_rate": 1.5325e-05,
+      "loss": 2.7054,
+      "step": 187
+    },
+    {
+      "epoch": 0.235,
+      "grad_norm": 1.5462883710861206,
+      "learning_rate": 1.5300000000000003e-05,
+      "loss": 3.0457,
+      "step": 188
+    },
+    {
+      "epoch": 0.23625,
+      "grad_norm": 2.16507625579834,
+      "learning_rate": 1.5275000000000002e-05,
+      "loss": 2.8668,
+      "step": 189
+    },
+    {
+      "epoch": 0.2375,
+      "grad_norm": 2.034597635269165,
+      "learning_rate": 1.525e-05,
+      "loss": 3.3775,
+      "step": 190
+    },
+    {
+      "epoch": 0.23875,
+      "grad_norm": 1.8364763259887695,
+      "learning_rate": 1.5225e-05,
+      "loss": 2.3838,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2459280490875244,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 2.604,
+      "step": 192
+    },
+    {
+      "epoch": 0.24125,
+      "grad_norm": 1.3169443607330322,
+      "learning_rate": 1.5175000000000001e-05,
+      "loss": 2.6639,
+      "step": 193
+    },
+    {
+      "epoch": 0.2425,
+      "grad_norm": 1.6183668375015259,
+      "learning_rate": 1.515e-05,
+      "loss": 2.663,
+      "step": 194
+    },
+    {
+      "epoch": 0.24375,
+      "grad_norm": 0.8612351417541504,
+      "learning_rate": 1.5125e-05,
+      "loss": 2.2988,
+      "step": 195
+    },
+    {
+      "epoch": 0.245,
+      "grad_norm": 1.1287574768066406,
+      "learning_rate": 1.5100000000000001e-05,
+      "loss": 2.5189,
+      "step": 196
+    },
+    {
+      "epoch": 0.24625,
+      "grad_norm": 0.8260729312896729,
+      "learning_rate": 1.5075000000000002e-05,
+      "loss": 2.1106,
+      "step": 197
+    },
+    {
+      "epoch": 0.2475,
+      "grad_norm": 1.1428359746932983,
+      "learning_rate": 1.505e-05,
+      "loss": 2.3685,
+      "step": 198
+    },
+    {
+      "epoch": 0.24875,
+      "grad_norm": 2.2927756309509277,
+      "learning_rate": 1.5025000000000001e-05,
+      "loss": 3.2356,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6453107595443726,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.9095,
+      "step": 200
+    },
+    {
+      "epoch": 0.25125,
+      "grad_norm": 2.4193787574768066,
+      "learning_rate": 1.4975000000000001e-05,
+      "loss": 3.0942,
+      "step": 201
+    },
+    {
+      "epoch": 0.2525,
+      "grad_norm": 1.536061406135559,
+      "learning_rate": 1.4950000000000003e-05,
+      "loss": 2.6016,
+      "step": 202
+    },
+    {
+      "epoch": 0.25375,
+      "grad_norm": 1.0485507249832153,
+      "learning_rate": 1.4925e-05,
+      "loss": 2.4134,
+      "step": 203
+    },
+    {
+      "epoch": 0.255,
+      "grad_norm": 2.519632577896118,
+      "learning_rate": 1.4900000000000001e-05,
+      "loss": 2.4384,
+      "step": 204
+    },
+    {
+      "epoch": 0.25625,
+      "grad_norm": 1.311072826385498,
+      "learning_rate": 1.4875000000000002e-05,
+      "loss": 3.0807,
+      "step": 205
+    },
+    {
+      "epoch": 0.2575,
+      "grad_norm": 2.9829020500183105,
+      "learning_rate": 1.4850000000000002e-05,
+      "loss": 3.4535,
+      "step": 206
+    },
+    {
+      "epoch": 0.25875,
+      "grad_norm": 1.2975739240646362,
+      "learning_rate": 1.4825000000000001e-05,
+      "loss": 2.5714,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9030929803848267,
+      "learning_rate": 1.48e-05,
+      "loss": 2.7733,
+      "step": 208
+    },
+    {
+      "epoch": 0.26125,
+      "grad_norm": 0.9590077996253967,
+      "learning_rate": 1.4775000000000002e-05,
+      "loss": 2.6125,
+      "step": 209
+    },
+    {
+      "epoch": 0.2625,
+      "grad_norm": 0.9263010025024414,
+      "learning_rate": 1.4750000000000003e-05,
+      "loss": 2.6456,
+      "step": 210
+    },
+    {
+      "epoch": 0.26375,
+      "grad_norm": 2.1573827266693115,
+      "learning_rate": 1.4725e-05,
+      "loss": 2.4447,
+      "step": 211
+    },
+    {
+      "epoch": 0.265,
+      "grad_norm": 1.386278510093689,
+      "learning_rate": 1.4700000000000002e-05,
+      "loss": 2.5971,
+      "step": 212
+    },
+    {
+      "epoch": 0.26625,
+      "grad_norm": 1.487113118171692,
+      "learning_rate": 1.4675000000000001e-05,
+      "loss": 2.8983,
+      "step": 213
+    },
+    {
+      "epoch": 0.2675,
+      "grad_norm": 1.6664327383041382,
+      "learning_rate": 1.4650000000000002e-05,
+      "loss": 2.0951,
+      "step": 214
+    },
+    {
+      "epoch": 0.26875,
+      "grad_norm": 1.2928472757339478,
+      "learning_rate": 1.4625e-05,
+      "loss": 2.7343,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1214581727981567,
+      "learning_rate": 1.46e-05,
+      "loss": 2.149,
+      "step": 216
+    },
+    {
+      "epoch": 0.27125,
+      "grad_norm": 1.724827766418457,
+      "learning_rate": 1.4575000000000002e-05,
+      "loss": 2.3816,
+      "step": 217
+    },
+    {
+      "epoch": 0.2725,
+      "grad_norm": 2.641063690185547,
+      "learning_rate": 1.4550000000000001e-05,
+      "loss": 3.357,
+      "step": 218
+    },
+    {
+      "epoch": 0.27375,
+      "grad_norm": 1.2942688465118408,
+      "learning_rate": 1.4525e-05,
+      "loss": 2.8008,
+      "step": 219
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 2.02486252784729,
+      "learning_rate": 1.45e-05,
+      "loss": 2.724,
+      "step": 220
+    },
+    {
+      "epoch": 0.27625,
+      "grad_norm": 2.8531992435455322,
+      "learning_rate": 1.4475000000000001e-05,
+      "loss": 2.8367,
+      "step": 221
+    },
+    {
+      "epoch": 0.2775,
+      "grad_norm": 0.9137434363365173,
+      "learning_rate": 1.4450000000000002e-05,
+      "loss": 2.0391,
+      "step": 222
+    },
+    {
+      "epoch": 0.27875,
+      "grad_norm": 2.9269864559173584,
+      "learning_rate": 1.4425e-05,
+      "loss": 2.9638,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.105649471282959,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 3.2201,
+      "step": 224
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 2.2428364753723145,
+      "learning_rate": 1.4375e-05,
+      "loss": 2.7664,
+      "step": 225
+    },
+    {
+      "epoch": 0.2825,
+      "grad_norm": 1.8540781736373901,
+      "learning_rate": 1.4350000000000002e-05,
+      "loss": 3.1237,
+      "step": 226
+    },
+    {
+      "epoch": 0.28375,
+      "grad_norm": 2.0040481090545654,
+      "learning_rate": 1.4325000000000003e-05,
+      "loss": 2.5248,
+      "step": 227
+    },
+    {
+      "epoch": 0.285,
+      "grad_norm": 1.4190040826797485,
+      "learning_rate": 1.43e-05,
+      "loss": 2.9788,
+      "step": 228
+    },
+    {
+      "epoch": 0.28625,
+      "grad_norm": 2.412309408187866,
+      "learning_rate": 1.4275000000000001e-05,
+      "loss": 2.568,
+      "step": 229
+    },
+    {
+      "epoch": 0.2875,
+      "grad_norm": 1.4534833431243896,
+      "learning_rate": 1.425e-05,
+      "loss": 2.7647,
+      "step": 230
+    },
+    {
+      "epoch": 0.28875,
+      "grad_norm": 0.662830650806427,
+      "learning_rate": 1.4225000000000002e-05,
+      "loss": 2.5177,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5043435096740723,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 2.8444,
+      "step": 232
+    },
+    {
+      "epoch": 0.29125,
+      "grad_norm": 1.3479336500167847,
+      "learning_rate": 1.4175e-05,
+      "loss": 2.2902,
+      "step": 233
+    },
+    {
+      "epoch": 0.2925,
+      "grad_norm": 1.9785624742507935,
+      "learning_rate": 1.4150000000000002e-05,
+      "loss": 2.8377,
+      "step": 234
+    },
+    {
+      "epoch": 0.29375,
+      "grad_norm": 3.683432102203369,
+      "learning_rate": 1.4125000000000003e-05,
+      "loss": 3.849,
+      "step": 235
+    },
+    {
+      "epoch": 0.295,
+      "grad_norm": 2.3539721965789795,
+      "learning_rate": 1.41e-05,
+      "loss": 3.0713,
+      "step": 236
+    },
+    {
+      "epoch": 0.29625,
+      "grad_norm": 2.2666311264038086,
+      "learning_rate": 1.4075000000000002e-05,
+      "loss": 2.6689,
+      "step": 237
+    },
+    {
+      "epoch": 0.2975,
+      "grad_norm": 0.7512000203132629,
+      "learning_rate": 1.4050000000000001e-05,
+      "loss": 2.1519,
+      "step": 238
+    },
+    {
+      "epoch": 0.29875,
+      "grad_norm": 1.755519986152649,
+      "learning_rate": 1.4025000000000002e-05,
+      "loss": 2.9841,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8116776943206787,
+      "learning_rate": 1.4e-05,
+      "loss": 2.4049,
+      "step": 240
+    },
+    {
+      "epoch": 0.30125,
+      "grad_norm": 1.5843620300292969,
+      "learning_rate": 1.3975000000000001e-05,
+      "loss": 2.7027,
+      "step": 241
+    },
+    {
+      "epoch": 0.3025,
+      "grad_norm": 1.8443893194198608,
+      "learning_rate": 1.3950000000000002e-05,
+      "loss": 2.6175,
+      "step": 242
+    },
+    {
+      "epoch": 0.30375,
+      "grad_norm": 4.015542030334473,
+      "learning_rate": 1.3925000000000001e-05,
+      "loss": 3.2365,
+      "step": 243
+    },
+    {
+      "epoch": 0.305,
+      "grad_norm": 1.515284776687622,
+      "learning_rate": 1.39e-05,
+      "loss": 3.4183,
+      "step": 244
+    },
+    {
+      "epoch": 0.30625,
+      "grad_norm": 1.4685897827148438,
+      "learning_rate": 1.3875e-05,
+      "loss": 2.5062,
+      "step": 245
+    },
+    {
+      "epoch": 0.3075,
+      "grad_norm": 1.9483778476715088,
+      "learning_rate": 1.3850000000000001e-05,
+      "loss": 2.6737,
+      "step": 246
+    },
+    {
+      "epoch": 0.30875,
+      "grad_norm": 2.194448471069336,
+      "learning_rate": 1.3825000000000002e-05,
+      "loss": 2.5334,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1393736600875854,
+      "learning_rate": 1.38e-05,
+      "loss": 2.452,
+      "step": 248
+    },
+    {
+      "epoch": 0.31125,
+      "grad_norm": 3.909176826477051,
+      "learning_rate": 1.3775000000000001e-05,
+      "loss": 3.8808,
+      "step": 249
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 1.2332671880722046,
+      "learning_rate": 1.375e-05,
+      "loss": 2.5233,
+      "step": 250
+    },
+    {
+      "epoch": 0.31375,
+      "grad_norm": 2.6342389583587646,
+      "learning_rate": 1.3725000000000002e-05,
+      "loss": 2.4018,
+      "step": 251
+    },
+    {
+      "epoch": 0.315,
+      "grad_norm": 2.0232162475585938,
+      "learning_rate": 1.3700000000000003e-05,
+      "loss": 2.704,
+      "step": 252
+    },
+    {
+      "epoch": 0.31625,
+      "grad_norm": 1.709433674812317,
+      "learning_rate": 1.3675e-05,
+      "loss": 3.0054,
+      "step": 253
+    },
+    {
+      "epoch": 0.3175,
+      "grad_norm": 1.425729751586914,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 2.5461,
+      "step": 254
+    },
+    {
+      "epoch": 0.31875,
+      "grad_norm": 1.902305245399475,
+      "learning_rate": 1.3625e-05,
+      "loss": 2.7863,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7936694622039795,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 2.5435,
+      "step": 256
+    },
+    {
+      "epoch": 0.32125,
+      "grad_norm": 2.072840929031372,
+      "learning_rate": 1.3575e-05,
+      "loss": 2.7425,
+      "step": 257
+    },
+    {
+      "epoch": 0.3225,
+      "grad_norm": 2.37286376953125,
+      "learning_rate": 1.355e-05,
+      "loss": 3.6158,
+      "step": 258
+    },
+    {
+      "epoch": 0.32375,
+      "grad_norm": 1.3893417119979858,
+      "learning_rate": 1.3525000000000002e-05,
+      "loss": 2.2764,
+      "step": 259
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 1.2564778327941895,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.9123,
+      "step": 260
+    },
+    {
+      "epoch": 0.32625,
+      "grad_norm": 1.7227541208267212,
+      "learning_rate": 1.3475e-05,
+      "loss": 2.8062,
+      "step": 261
+    },
+    {
+      "epoch": 0.3275,
+      "grad_norm": 1.5551847219467163,
+      "learning_rate": 1.3450000000000002e-05,
+      "loss": 2.5495,
+      "step": 262
+    },
+    {
+      "epoch": 0.32875,
+      "grad_norm": 3.329219102859497,
+      "learning_rate": 1.3425000000000001e-05,
+      "loss": 3.1482,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6194255352020264,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 2.0326,
+      "step": 264
+    },
+    {
+      "epoch": 0.33125,
+      "grad_norm": 2.509632110595703,
+      "learning_rate": 1.3375e-05,
+      "loss": 2.49,
+      "step": 265
+    },
+    {
+      "epoch": 0.3325,
+      "grad_norm": 1.3284372091293335,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 2.7824,
+      "step": 266
+    },
+    {
+      "epoch": 0.33375,
+      "grad_norm": 1.663150429725647,
+      "learning_rate": 1.3325000000000002e-05,
+      "loss": 2.5316,
+      "step": 267
+    },
+    {
+      "epoch": 0.335,
+      "grad_norm": 1.093685507774353,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 2.3014,
+      "step": 268
+    },
+    {
+      "epoch": 0.33625,
+      "grad_norm": 2.4130373001098633,
+      "learning_rate": 1.3275e-05,
+      "loss": 2.3933,
+      "step": 269
+    },
+    {
+      "epoch": 0.3375,
+      "grad_norm": 1.6137781143188477,
+      "learning_rate": 1.325e-05,
+      "loss": 1.8412,
+      "step": 270
+    },
+    {
+      "epoch": 0.33875,
+      "grad_norm": 2.0770466327667236,
+      "learning_rate": 1.3225000000000001e-05,
+      "loss": 2.4478,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.511610269546509,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 2.6513,
+      "step": 272
+    },
+    {
+      "epoch": 0.34125,
+      "grad_norm": 1.3404005765914917,
+      "learning_rate": 1.3175e-05,
+      "loss": 2.3267,
+      "step": 273
+    },
+    {
+      "epoch": 0.3425,
+      "grad_norm": 2.10129714012146,
+      "learning_rate": 1.3150000000000001e-05,
+      "loss": 2.8609,
+      "step": 274
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 4.970488548278809,
+      "learning_rate": 1.3125e-05,
+      "loss": 4.5557,
+      "step": 275
+    },
+    {
+      "epoch": 0.345,
+      "grad_norm": 1.0287278890609741,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 1.9752,
+      "step": 276
+    },
+    {
+      "epoch": 0.34625,
+      "grad_norm": 1.4155839681625366,
+      "learning_rate": 1.3075000000000003e-05,
+      "loss": 2.5988,
+      "step": 277
+    },
+    {
+      "epoch": 0.3475,
+      "grad_norm": 2.1226813793182373,
+      "learning_rate": 1.305e-05,
+      "loss": 2.4375,
+      "step": 278
+    },
+    {
+      "epoch": 0.34875,
+      "grad_norm": 1.3161485195159912,
+      "learning_rate": 1.3025000000000002e-05,
+      "loss": 2.3141,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0883145332336426,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.2943,
+      "step": 280
+    },
+    {
+      "epoch": 0.35125,
+      "grad_norm": 0.6219211220741272,
+      "learning_rate": 1.2975000000000002e-05,
+      "loss": 2.7156,
+      "step": 281
+    },
+    {
+      "epoch": 0.3525,
+      "grad_norm": 3.4014060497283936,
+      "learning_rate": 1.295e-05,
+      "loss": 1.9477,
+      "step": 282
+    },
+    {
+      "epoch": 0.35375,
+      "grad_norm": 1.447366714477539,
+      "learning_rate": 1.2925e-05,
+      "loss": 3.2115,
+      "step": 283
+    },
+    {
+      "epoch": 0.355,
+      "grad_norm": 0.8695298433303833,
+      "learning_rate": 1.2900000000000002e-05,
+      "loss": 2.0062,
+      "step": 284
+    },
+    {
+      "epoch": 0.35625,
+      "grad_norm": 1.2442666292190552,
+      "learning_rate": 1.2875000000000001e-05,
+      "loss": 2.782,
+      "step": 285
+    },
+    {
+      "epoch": 0.3575,
+      "grad_norm": 4.334051609039307,
+      "learning_rate": 1.285e-05,
+      "loss": 3.7922,
+      "step": 286
+    },
+    {
+      "epoch": 0.35875,
+      "grad_norm": 1.0075643062591553,
+      "learning_rate": 1.2825e-05,
+      "loss": 2.3886,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.952709436416626,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 3.1041,
+      "step": 288
+    },
+    {
+      "epoch": 0.36125,
+      "grad_norm": 1.9426610469818115,
+      "learning_rate": 1.2775000000000002e-05,
+      "loss": 2.5739,
+      "step": 289
+    },
+    {
+      "epoch": 0.3625,
+      "grad_norm": 1.2706924676895142,
+      "learning_rate": 1.275e-05,
+      "loss": 2.6037,
+      "step": 290
+    },
+    {
+      "epoch": 0.36375,
+      "grad_norm": 1.1551876068115234,
+      "learning_rate": 1.2725000000000001e-05,
+      "loss": 2.2662,
+      "step": 291
+    },
+    {
+      "epoch": 0.365,
+      "grad_norm": 2.400639772415161,
+      "learning_rate": 1.27e-05,
+      "loss": 3.0005,
+      "step": 292
+    },
+    {
+      "epoch": 0.36625,
+      "grad_norm": 2.8095648288726807,
+      "learning_rate": 1.2675000000000001e-05,
+      "loss": 3.0654,
+      "step": 293
+    },
+    {
+      "epoch": 0.3675,
+      "grad_norm": 0.8239273428916931,
+      "learning_rate": 1.2650000000000001e-05,
+      "loss": 1.9067,
+      "step": 294
+    },
+    {
+      "epoch": 0.36875,
+      "grad_norm": 1.5768872499465942,
+      "learning_rate": 1.2625e-05,
+      "loss": 2.4867,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.463496208190918,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 3.0372,
+      "step": 296
+    },
+    {
+      "epoch": 0.37125,
+      "grad_norm": 1.8864333629608154,
+      "learning_rate": 1.2575000000000002e-05,
+      "loss": 2.5445,
+      "step": 297
+    },
+    {
+      "epoch": 0.3725,
+      "grad_norm": 2.185873508453369,
+      "learning_rate": 1.255e-05,
+      "loss": 2.7776,
+      "step": 298
+    },
+    {
+      "epoch": 0.37375,
+      "grad_norm": 1.242254376411438,
+      "learning_rate": 1.2525000000000001e-05,
+      "loss": 2.6274,
+      "step": 299
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 1.5784997940063477,
+      "learning_rate": 1.25e-05,
+      "loss": 2.6328,
+      "step": 300
+    },
+    {
+      "epoch": 0.37625,
+      "grad_norm": 2.0619335174560547,
+      "learning_rate": 1.2475000000000002e-05,
+      "loss": 2.2983,
+      "step": 301
+    },
+    {
+      "epoch": 0.3775,
+      "grad_norm": 1.7328943014144897,
+      "learning_rate": 1.2450000000000003e-05,
+      "loss": 2.3851,
+      "step": 302
+    },
+    {
+      "epoch": 0.37875,
+      "grad_norm": 1.7219791412353516,
+      "learning_rate": 1.2425e-05,
+      "loss": 2.1684,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9850049018859863,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 3.1503,
+      "step": 304
+    },
+    {
+      "epoch": 0.38125,
+      "grad_norm": 1.193284511566162,
+      "learning_rate": 1.2375000000000001e-05,
+      "loss": 2.8272,
+      "step": 305
+    },
+    {
+      "epoch": 0.3825,
+      "grad_norm": 8.433061599731445,
+      "learning_rate": 1.2350000000000002e-05,
+      "loss": 2.7003,
+      "step": 306
+    },
+    {
+      "epoch": 0.38375,
+      "grad_norm": 1.3799391984939575,
+      "learning_rate": 1.2325e-05,
+      "loss": 2.7715,
+      "step": 307
+    },
+    {
+      "epoch": 0.385,
+      "grad_norm": 1.137082576751709,
+      "learning_rate": 1.23e-05,
+      "loss": 2.6402,
+      "step": 308
+    },
+    {
+      "epoch": 0.38625,
+      "grad_norm": 3.6281142234802246,
+      "learning_rate": 1.2275000000000002e-05,
+      "loss": 3.1791,
+      "step": 309
+    },
+    {
+      "epoch": 0.3875,
+      "grad_norm": 1.4617842435836792,
+      "learning_rate": 1.2250000000000001e-05,
+      "loss": 2.2238,
+      "step": 310
+    },
+    {
+      "epoch": 0.38875,
+      "grad_norm": 1.729715347290039,
+      "learning_rate": 1.2225e-05,
+      "loss": 3.2183,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7599000930786133,
+      "learning_rate": 1.22e-05,
+      "loss": 2.6337,
+      "step": 312
+    },
+    {
+      "epoch": 0.39125,
+      "grad_norm": 2.0601325035095215,
+      "learning_rate": 1.2175000000000001e-05,
+      "loss": 2.7048,
+      "step": 313
+    },
+    {
+      "epoch": 0.3925,
+      "grad_norm": 1.496025562286377,
+      "learning_rate": 1.2150000000000002e-05,
+      "loss": 2.2228,
+      "step": 314
+    },
+    {
+      "epoch": 0.39375,
+      "grad_norm": 2.1809468269348145,
+      "learning_rate": 1.2125e-05,
+      "loss": 2.1678,
+      "step": 315
+    },
+    {
+      "epoch": 0.395,
+      "grad_norm": 1.5901917219161987,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 1.8807,
+      "step": 316
+    },
+    {
+      "epoch": 0.39625,
+      "grad_norm": 1.4371182918548584,
+      "learning_rate": 1.2075e-05,
+      "loss": 2.0627,
+      "step": 317
+    },
+    {
+      "epoch": 0.3975,
+      "grad_norm": 1.4142842292785645,
+      "learning_rate": 1.2050000000000002e-05,
+      "loss": 1.93,
+      "step": 318
+    },
+    {
+      "epoch": 0.39875,
+      "grad_norm": 1.0198673009872437,
+      "learning_rate": 1.2025e-05,
+      "loss": 2.3295,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1797082424163818,
+      "learning_rate": 1.2e-05,
+      "loss": 2.5342,
+      "step": 320
+    },
+    {
+      "epoch": 0.40125,
+      "grad_norm": 1.0422452688217163,
+      "learning_rate": 1.1975000000000001e-05,
+      "loss": 2.0141,
+      "step": 321
+    },
+    {
+      "epoch": 0.4025,
+      "grad_norm": 1.1430671215057373,
+      "learning_rate": 1.195e-05,
+      "loss": 1.8992,
+      "step": 322
+    },
+    {
+      "epoch": 0.40375,
+      "grad_norm": 2.177053928375244,
+      "learning_rate": 1.1925e-05,
+      "loss": 2.7447,
+      "step": 323
+    },
+    {
+      "epoch": 0.405,
+      "grad_norm": 2.095257520675659,
+      "learning_rate": 1.1900000000000001e-05,
+      "loss": 2.5267,
+      "step": 324
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.4478448629379272,
+      "learning_rate": 1.1875e-05,
+      "loss": 2.6989,
+      "step": 325
+    },
+    {
+      "epoch": 0.4075,
+      "grad_norm": 1.9374433755874634,
+      "learning_rate": 1.1850000000000002e-05,
+      "loss": 2.4978,
+      "step": 326
+    },
+    {
+      "epoch": 0.40875,
+      "grad_norm": 0.9773975014686584,
+      "learning_rate": 1.1825000000000003e-05,
+      "loss": 2.6889,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5445204973220825,
+      "learning_rate": 1.18e-05,
+      "loss": 2.4233,
+      "step": 328
+    },
+    {
+      "epoch": 0.41125,
+      "grad_norm": 1.492608904838562,
+      "learning_rate": 1.1775000000000002e-05,
+      "loss": 2.6976,
+      "step": 329
+    },
+    {
+      "epoch": 0.4125,
+      "grad_norm": 2.212707996368408,
+      "learning_rate": 1.1750000000000001e-05,
+      "loss": 3.2724,
+      "step": 330
+    },
+    {
+      "epoch": 0.41375,
+      "grad_norm": 1.4107556343078613,
+      "learning_rate": 1.1725000000000002e-05,
+      "loss": 3.0866,
+      "step": 331
+    },
+    {
+      "epoch": 0.415,
+      "grad_norm": 2.913604736328125,
+      "learning_rate": 1.17e-05,
+      "loss": 2.4357,
+      "step": 332
+    },
+    {
+      "epoch": 0.41625,
+      "grad_norm": 3.2254488468170166,
+      "learning_rate": 1.1675000000000001e-05,
+      "loss": 3.5358,
+      "step": 333
+    },
+    {
+      "epoch": 0.4175,
+      "grad_norm": 1.6059589385986328,
+      "learning_rate": 1.1650000000000002e-05,
+      "loss": 2.6456,
+      "step": 334
+    },
+    {
+      "epoch": 0.41875,
+      "grad_norm": 2.4393796920776367,
+      "learning_rate": 1.1625000000000001e-05,
+      "loss": 2.0989,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5735288858413696,
+      "learning_rate": 1.16e-05,
+      "loss": 2.979,
+      "step": 336
+    },
+    {
+      "epoch": 0.42125,
+      "grad_norm": 1.5758427381515503,
+      "learning_rate": 1.1575e-05,
+      "loss": 2.7381,
+      "step": 337
+    },
+    {
+      "epoch": 0.4225,
+      "grad_norm": 0.8109416961669922,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 2.3608,
+      "step": 338
+    },
+    {
+      "epoch": 0.42375,
+      "grad_norm": 1.8177721500396729,
+      "learning_rate": 1.1525000000000002e-05,
+      "loss": 2.8134,
+      "step": 339
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 1.5914677381515503,
+      "learning_rate": 1.15e-05,
+      "loss": 2.4923,
+      "step": 340
+    },
+    {
+      "epoch": 0.42625,
+      "grad_norm": 0.7988778352737427,
+      "learning_rate": 1.1475000000000001e-05,
+      "loss": 1.855,
+      "step": 341
+    },
+    {
+      "epoch": 0.4275,
+      "grad_norm": 1.4848300218582153,
+      "learning_rate": 1.145e-05,
+      "loss": 2.1767,
+      "step": 342
+    },
+    {
+      "epoch": 0.42875,
+      "grad_norm": 2.404296875,
+      "learning_rate": 1.1425000000000002e-05,
+      "loss": 2.5743,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8775387406349182,
+      "learning_rate": 1.14e-05,
+      "loss": 2.0211,
+      "step": 344
+    },
+    {
+      "epoch": 0.43125,
+      "grad_norm": 1.3193469047546387,
+      "learning_rate": 1.1375e-05,
+      "loss": 1.3909,
+      "step": 345
+    },
+    {
+      "epoch": 0.4325,
+      "grad_norm": 2.3813071250915527,
+      "learning_rate": 1.1350000000000001e-05,
+      "loss": 3.1726,
+      "step": 346
+    },
+    {
+      "epoch": 0.43375,
+      "grad_norm": 1.7489192485809326,
+      "learning_rate": 1.1325e-05,
+      "loss": 2.557,
+      "step": 347
+    },
+    {
+      "epoch": 0.435,
+      "grad_norm": 1.4432573318481445,
+      "learning_rate": 1.13e-05,
+      "loss": 2.8275,
+      "step": 348
+    },
+    {
+      "epoch": 0.43625,
+      "grad_norm": 1.5648714303970337,
+      "learning_rate": 1.1275e-05,
+      "loss": 1.975,
+      "step": 349
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 1.1547632217407227,
+      "learning_rate": 1.125e-05,
+      "loss": 1.8893,
+      "step": 350
+    },
+    {
+      "epoch": 0.43875,
+      "grad_norm": 1.3992825746536255,
+      "learning_rate": 1.1225000000000002e-05,
+      "loss": 2.3774,
+      "step": 351
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0416218042373657,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 2.3281,
+      "step": 352
+    },
+    {
+      "epoch": 0.44125,
+      "grad_norm": 1.459821105003357,
+      "learning_rate": 1.1175e-05,
+      "loss": 2.1034,
+      "step": 353
+    },
+    {
+      "epoch": 0.4425,
+      "grad_norm": 1.4830873012542725,
+      "learning_rate": 1.1150000000000002e-05,
+      "loss": 2.8088,
+      "step": 354
+    },
+    {
+      "epoch": 0.44375,
+      "grad_norm": 1.200899362564087,
+      "learning_rate": 1.1125000000000001e-05,
+      "loss": 2.3231,
+      "step": 355
+    },
+    {
+      "epoch": 0.445,
+      "grad_norm": 2.005309820175171,
+      "learning_rate": 1.1100000000000002e-05,
+      "loss": 3.5848,
+      "step": 356
+    },
+    {
+      "epoch": 0.44625,
+      "grad_norm": 1.3485474586486816,
+      "learning_rate": 1.1075e-05,
+      "loss": 2.7273,
+      "step": 357
+    },
+    {
+      "epoch": 0.4475,
+      "grad_norm": 1.5020636320114136,
+      "learning_rate": 1.1050000000000001e-05,
+      "loss": 2.5139,
+      "step": 358
+    },
+    {
+      "epoch": 0.44875,
+      "grad_norm": 1.4698582887649536,
+      "learning_rate": 1.1025000000000002e-05,
+      "loss": 2.4339,
+      "step": 359
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5619611740112305,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.958,
+      "step": 360
+    },
+    {
+      "epoch": 0.45125,
+      "grad_norm": 1.1998145580291748,
+      "learning_rate": 1.0975e-05,
+      "loss": 2.4375,
+      "step": 361
+    },
+    {
+      "epoch": 0.4525,
+      "grad_norm": 3.1132500171661377,
+      "learning_rate": 1.095e-05,
+      "loss": 3.1062,
+      "step": 362
+    },
+    {
+      "epoch": 0.45375,
+      "grad_norm": 3.445772886276245,
+      "learning_rate": 1.0925000000000001e-05,
+      "loss": 3.3763,
+      "step": 363
+    },
+    {
+      "epoch": 0.455,
+      "grad_norm": 2.5865681171417236,
+      "learning_rate": 1.0900000000000002e-05,
+      "loss": 2.6878,
+      "step": 364
+    },
+    {
+      "epoch": 0.45625,
+      "grad_norm": 2.759075880050659,
+      "learning_rate": 1.0875e-05,
+      "loss": 2.9461,
+      "step": 365
+    },
+    {
+      "epoch": 0.4575,
+      "grad_norm": 2.7135589122772217,
+      "learning_rate": 1.0850000000000001e-05,
+      "loss": 2.7731,
+      "step": 366
+    },
+    {
+      "epoch": 0.45875,
+      "grad_norm": 3.1075406074523926,
+      "learning_rate": 1.0825e-05,
+      "loss": 2.9753,
+      "step": 367
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8841419219970703,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 2.4276,
+      "step": 368
+    },
+    {
+      "epoch": 0.46125,
+      "grad_norm": 1.5538018941879272,
+      "learning_rate": 1.0775e-05,
+      "loss": 2.5934,
+      "step": 369
+    },
+    {
+      "epoch": 0.4625,
+      "grad_norm": 1.6202460527420044,
+      "learning_rate": 1.075e-05,
+      "loss": 2.2371,
+      "step": 370
+    },
+    {
+      "epoch": 0.46375,
+      "grad_norm": 1.8580071926116943,
+      "learning_rate": 1.0725000000000001e-05,
+      "loss": 2.7594,
+      "step": 371
+    },
+    {
+      "epoch": 0.465,
+      "grad_norm": 1.1591219902038574,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 2.0254,
+      "step": 372
+    },
+    {
+      "epoch": 0.46625,
+      "grad_norm": 1.995326042175293,
+      "learning_rate": 1.0675e-05,
+      "loss": 2.7667,
+      "step": 373
+    },
+    {
+      "epoch": 0.4675,
+      "grad_norm": 1.5032466650009155,
+      "learning_rate": 1.065e-05,
+      "loss": 2.5207,
+      "step": 374
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 2.4473445415496826,
+      "learning_rate": 1.0625e-05,
+      "loss": 2.9158,
+      "step": 375
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.215508460998535,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 3.2581,
+      "step": 376
+    },
+    {
+      "epoch": 0.47125,
+      "grad_norm": 1.3192355632781982,
+      "learning_rate": 1.0575000000000001e-05,
+      "loss": 2.4273,
+      "step": 377
+    },
+    {
+      "epoch": 0.4725,
+      "grad_norm": 1.9035662412643433,
+      "learning_rate": 1.055e-05,
+      "loss": 2.9935,
+      "step": 378
+    },
+    {
+      "epoch": 0.47375,
+      "grad_norm": 1.2860153913497925,
+      "learning_rate": 1.0525e-05,
+      "loss": 2.6538,
+      "step": 379
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 1.6805506944656372,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 2.5256,
+      "step": 380
+    },
+    {
+      "epoch": 0.47625,
+      "grad_norm": 1.1306779384613037,
+      "learning_rate": 1.0475000000000002e-05,
+      "loss": 2.124,
+      "step": 381
+    },
+    {
+      "epoch": 0.4775,
+      "grad_norm": 2.3374178409576416,
+      "learning_rate": 1.045e-05,
+      "loss": 2.8561,
+      "step": 382
+    },
+    {
+      "epoch": 0.47875,
+      "grad_norm": 1.739198088645935,
+      "learning_rate": 1.0425000000000001e-05,
+      "loss": 2.6378,
+      "step": 383
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.08516526222229,
+      "learning_rate": 1.04e-05,
+      "loss": 3.1051,
+      "step": 384
+    },
+    {
+      "epoch": 0.48125,
+      "grad_norm": 1.6211674213409424,
+      "learning_rate": 1.0375000000000001e-05,
+      "loss": 3.189,
+      "step": 385
+    },
+    {
+      "epoch": 0.4825,
+      "grad_norm": 1.768182396888733,
+      "learning_rate": 1.0350000000000001e-05,
+      "loss": 2.268,
+      "step": 386
+    },
+    {
+      "epoch": 0.48375,
+      "grad_norm": 1.2901172637939453,
+      "learning_rate": 1.0325e-05,
+      "loss": 2.2684,
+      "step": 387
+    },
+    {
+      "epoch": 0.485,
+      "grad_norm": 1.6452449560165405,
+      "learning_rate": 1.0300000000000001e-05,
+      "loss": 2.7506,
+      "step": 388
+    },
+    {
+      "epoch": 0.48625,
+      "grad_norm": 1.1954340934753418,
+      "learning_rate": 1.0275000000000002e-05,
+      "loss": 2.1034,
+      "step": 389
+    },
+    {
+      "epoch": 0.4875,
+      "grad_norm": 1.6171998977661133,
+      "learning_rate": 1.025e-05,
+      "loss": 2.6603,
+      "step": 390
+    },
+    {
+      "epoch": 0.48875,
+      "grad_norm": 1.2973988056182861,
+      "learning_rate": 1.0225000000000001e-05,
+      "loss": 2.6099,
+      "step": 391
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1014139652252197,
+      "learning_rate": 1.02e-05,
+      "loss": 2.9254,
+      "step": 392
+    },
+    {
+      "epoch": 0.49125,
+      "grad_norm": 1.9552756547927856,
+      "learning_rate": 1.0175000000000002e-05,
+      "loss": 2.892,
+      "step": 393
+    },
+    {
+      "epoch": 0.4925,
+      "grad_norm": 2.1882715225219727,
+      "learning_rate": 1.015e-05,
+      "loss": 3.0698,
+      "step": 394
+    },
+    {
+      "epoch": 0.49375,
+      "grad_norm": 1.013117790222168,
+      "learning_rate": 1.0125e-05,
+      "loss": 2.7269,
+      "step": 395
+    },
+    {
+      "epoch": 0.495,
+      "grad_norm": 1.3441874980926514,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 1.7805,
+      "step": 396
+    },
+    {
+      "epoch": 0.49625,
+      "grad_norm": 1.674667477607727,
+      "learning_rate": 1.0075000000000001e-05,
+      "loss": 2.3934,
+      "step": 397
+    },
+    {
+      "epoch": 0.4975,
+      "grad_norm": 3.498429536819458,
+      "learning_rate": 1.005e-05,
+      "loss": 2.9463,
+      "step": 398
+    },
+    {
+      "epoch": 0.49875,
+      "grad_norm": 1.2237039804458618,
+      "learning_rate": 1.0025e-05,
+      "loss": 2.7709,
+      "step": 399
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.075360059738159,
+      "learning_rate": 1e-05,
+      "loss": 3.304,
+      "step": 400
+    },
+    {
+      "epoch": 0.50125,
+      "grad_norm": 1.4542925357818604,
+      "learning_rate": 9.975000000000002e-06,
+      "loss": 2.1337,
+      "step": 401
+    },
+    {
+      "epoch": 0.5025,
+      "grad_norm": 2.508572816848755,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 2.842,
+      "step": 402
+    },
+    {
+      "epoch": 0.50375,
+      "grad_norm": 2.260857105255127,
+      "learning_rate": 9.925e-06,
+      "loss": 2.6903,
+      "step": 403
+    },
+    {
+      "epoch": 0.505,
+      "grad_norm": 1.9771928787231445,
+      "learning_rate": 9.9e-06,
+      "loss": 2.4159,
+      "step": 404
+    },
+    {
+      "epoch": 0.50625,
+      "grad_norm": 1.307166337966919,
+      "learning_rate": 9.875000000000001e-06,
+      "loss": 2.2289,
+      "step": 405
+    },
+    {
+      "epoch": 0.5075,
+      "grad_norm": 2.5441768169403076,
+      "learning_rate": 9.85e-06,
+      "loss": 2.4819,
+      "step": 406
+    },
+    {
+      "epoch": 0.50875,
+      "grad_norm": 1.579392433166504,
+      "learning_rate": 9.825000000000002e-06,
+      "loss": 2.5208,
+      "step": 407
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8270081281661987,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.2591,
+      "step": 408
+    },
+    {
+      "epoch": 0.51125,
+      "grad_norm": 1.8742762804031372,
+      "learning_rate": 9.775e-06,
+      "loss": 2.3944,
+      "step": 409
+    },
+    {
+      "epoch": 0.5125,
+      "grad_norm": 1.6326040029525757,
+      "learning_rate": 9.75e-06,
+      "loss": 2.7157,
+      "step": 410
+    },
+    {
+      "epoch": 0.51375,
+      "grad_norm": 2.675300121307373,
+      "learning_rate": 9.725000000000001e-06,
+      "loss": 2.648,
+      "step": 411
+    },
+    {
+      "epoch": 0.515,
+      "grad_norm": 1.707234263420105,
+      "learning_rate": 9.7e-06,
+      "loss": 2.2568,
+      "step": 412
+    },
+    {
+      "epoch": 0.51625,
+      "grad_norm": 1.860190987586975,
+      "learning_rate": 9.675000000000001e-06,
+      "loss": 2.3879,
+      "step": 413
+    },
+    {
+      "epoch": 0.5175,
+      "grad_norm": 2.554323673248291,
+      "learning_rate": 9.65e-06,
+      "loss": 2.4429,
+      "step": 414
+    },
+    {
+      "epoch": 0.51875,
+      "grad_norm": 3.1000308990478516,
+      "learning_rate": 9.625e-06,
+      "loss": 3.6857,
+      "step": 415
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.2858200073242188,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.3423,
+      "step": 416
+    },
+    {
+      "epoch": 0.52125,
+      "grad_norm": 1.7305254936218262,
+      "learning_rate": 9.575e-06,
+      "loss": 2.5126,
+      "step": 417
+    },
+    {
+      "epoch": 0.5225,
+      "grad_norm": 2.0434417724609375,
+      "learning_rate": 9.55e-06,
+      "loss": 2.8066,
+      "step": 418
+    },
+    {
+      "epoch": 0.52375,
+      "grad_norm": 1.1936882734298706,
+      "learning_rate": 9.525000000000001e-06,
+      "loss": 1.9428,
+      "step": 419
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 1.5286387205123901,
+      "learning_rate": 9.5e-06,
+      "loss": 2.0241,
+      "step": 420
+    },
+    {
+      "epoch": 0.52625,
+      "grad_norm": 1.162139654159546,
+      "learning_rate": 9.475000000000002e-06,
+      "loss": 2.8492,
+      "step": 421
+    },
+    {
+      "epoch": 0.5275,
+      "grad_norm": 2.5677266120910645,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 1.8682,
+      "step": 422
+    },
+    {
+      "epoch": 0.52875,
+      "grad_norm": 2.4863381385803223,
+      "learning_rate": 9.425e-06,
+      "loss": 2.8088,
+      "step": 423
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.074288845062256,
+      "learning_rate": 9.4e-06,
+      "loss": 3.1033,
+      "step": 424
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 3.1781013011932373,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 2.7687,
+      "step": 425
+    },
+    {
+      "epoch": 0.5325,
+      "grad_norm": 1.8094357252120972,
+      "learning_rate": 9.350000000000002e-06,
+      "loss": 2.6157,
+      "step": 426
+    },
+    {
+      "epoch": 0.53375,
+      "grad_norm": 1.5501965284347534,
+      "learning_rate": 9.325000000000001e-06,
+      "loss": 2.5792,
+      "step": 427
+    },
+    {
+      "epoch": 0.535,
+      "grad_norm": 2.8156535625457764,
+      "learning_rate": 9.3e-06,
+      "loss": 2.4914,
+      "step": 428
+    },
+    {
+      "epoch": 0.53625,
+      "grad_norm": 1.1889698505401611,
+      "learning_rate": 9.275e-06,
+      "loss": 2.4051,
+      "step": 429
+    },
+    {
+      "epoch": 0.5375,
+      "grad_norm": 2.8830533027648926,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 2.5135,
+      "step": 430
+    },
+    {
+      "epoch": 0.53875,
+      "grad_norm": 1.8553954362869263,
+      "learning_rate": 9.225e-06,
+      "loss": 3.0061,
+      "step": 431
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.38657283782959,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.3366,
+      "step": 432
+    },
+    {
+      "epoch": 0.54125,
+      "grad_norm": 2.5273995399475098,
+      "learning_rate": 9.175000000000001e-06,
+      "loss": 2.3687,
+      "step": 433
+    },
+    {
+      "epoch": 0.5425,
+      "grad_norm": 2.155266761779785,
+      "learning_rate": 9.15e-06,
+      "loss": 3.3555,
+      "step": 434
+    },
+    {
+      "epoch": 0.54375,
+      "grad_norm": 2.964810371398926,
+      "learning_rate": 9.125e-06,
+      "loss": 3.2155,
+      "step": 435
+    },
+    {
+      "epoch": 0.545,
+      "grad_norm": 1.2716864347457886,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.637,
+      "step": 436
+    },
+    {
+      "epoch": 0.54625,
+      "grad_norm": 1.1982064247131348,
+      "learning_rate": 9.075e-06,
+      "loss": 2.5646,
+      "step": 437
+    },
+    {
+      "epoch": 0.5475,
+      "grad_norm": 3.224846601486206,
+      "learning_rate": 9.050000000000001e-06,
+      "loss": 2.0353,
+      "step": 438
+    },
+    {
+      "epoch": 0.54875,
+      "grad_norm": 2.4795050621032715,
+      "learning_rate": 9.025e-06,
+      "loss": 2.4973,
+      "step": 439
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.392571210861206,
+      "learning_rate": 9e-06,
+      "loss": 1.8671,
+      "step": 440
+    },
+    {
+      "epoch": 0.55125,
+      "grad_norm": 2.579519748687744,
+      "learning_rate": 8.975e-06,
+      "loss": 2.4024,
+      "step": 441
+    },
+    {
+      "epoch": 0.5525,
+      "grad_norm": 1.491835594177246,
+      "learning_rate": 8.95e-06,
+      "loss": 2.8333,
+      "step": 442
+    },
+    {
+      "epoch": 0.55375,
+      "grad_norm": 1.6103180646896362,
+      "learning_rate": 8.925e-06,
+      "loss": 2.4064,
+      "step": 443
+    },
+    {
+      "epoch": 0.555,
+      "grad_norm": 1.8389121294021606,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.8206,
+      "step": 444
+    },
+    {
+      "epoch": 0.55625,
+      "grad_norm": 1.8887784481048584,
+      "learning_rate": 8.875e-06,
+      "loss": 3.0003,
+      "step": 445
+    },
+    {
+      "epoch": 0.5575,
+      "grad_norm": 3.0405466556549072,
+      "learning_rate": 8.85e-06,
+      "loss": 3.1721,
+      "step": 446
+    },
+    {
+      "epoch": 0.55875,
+      "grad_norm": 1.5504568815231323,
+      "learning_rate": 8.825000000000001e-06,
+      "loss": 3.1523,
+      "step": 447
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4831838607788086,
+      "learning_rate": 8.8e-06,
+      "loss": 2.9069,
+      "step": 448
+    },
+    {
+      "epoch": 0.56125,
+      "grad_norm": 1.7960699796676636,
+      "learning_rate": 8.775e-06,
+      "loss": 2.3023,
+      "step": 449
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 2.359325885772705,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.7974,
+      "step": 450
+    },
+    {
+      "epoch": 0.56375,
+      "grad_norm": 3.2360641956329346,
+      "learning_rate": 8.725000000000002e-06,
+      "loss": 3.0734,
+      "step": 451
+    },
+    {
+      "epoch": 0.565,
+      "grad_norm": 1.5362831354141235,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.1028,
+      "step": 452
+    },
+    {
+      "epoch": 0.56625,
+      "grad_norm": 0.892631471157074,
+      "learning_rate": 8.675e-06,
+      "loss": 2.0199,
+      "step": 453
+    },
+    {
+      "epoch": 0.5675,
+      "grad_norm": 2.0016226768493652,
+      "learning_rate": 8.65e-06,
+      "loss": 2.5738,
+      "step": 454
+    },
+    {
+      "epoch": 0.56875,
+      "grad_norm": 1.3834383487701416,
+      "learning_rate": 8.625000000000001e-06,
+      "loss": 2.8472,
+      "step": 455
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3233683109283447,
+      "learning_rate": 8.6e-06,
+      "loss": 2.653,
+      "step": 456
+    },
+    {
+      "epoch": 0.57125,
+      "grad_norm": 1.5941835641860962,
+      "learning_rate": 8.575000000000002e-06,
+      "loss": 3.1008,
+      "step": 457
+    },
+    {
+      "epoch": 0.5725,
+      "grad_norm": 1.9550966024398804,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 2.1212,
+      "step": 458
+    },
+    {
+      "epoch": 0.57375,
+      "grad_norm": 1.5184082984924316,
+      "learning_rate": 8.525e-06,
+      "loss": 2.6667,
+      "step": 459
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 0.9546151757240295,
+      "learning_rate": 8.5e-06,
+      "loss": 2.6412,
+      "step": 460
+    },
+    {
+      "epoch": 0.57625,
+      "grad_norm": 1.303501844406128,
+      "learning_rate": 8.475000000000001e-06,
+      "loss": 2.4247,
+      "step": 461
+    },
+    {
+      "epoch": 0.5775,
+      "grad_norm": 1.3831559419631958,
+      "learning_rate": 8.45e-06,
+      "loss": 2.315,
+      "step": 462
+    },
+    {
+      "epoch": 0.57875,
+      "grad_norm": 1.7408896684646606,
+      "learning_rate": 8.425000000000001e-06,
+      "loss": 2.9685,
+      "step": 463
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8592466115951538,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.3298,
+      "step": 464
+    },
+    {
+      "epoch": 0.58125,
+      "grad_norm": 1.8565177917480469,
+      "learning_rate": 8.375e-06,
+      "loss": 3.0411,
+      "step": 465
+    },
+    {
+      "epoch": 0.5825,
+      "grad_norm": 1.3421576023101807,
+      "learning_rate": 8.35e-06,
+      "loss": 3.0402,
+      "step": 466
+    },
+    {
+      "epoch": 0.58375,
+      "grad_norm": 2.391731023788452,
+      "learning_rate": 8.325e-06,
+      "loss": 2.8541,
+      "step": 467
+    },
+    {
+      "epoch": 0.585,
+      "grad_norm": 1.367205023765564,
+      "learning_rate": 8.3e-06,
+      "loss": 2.5367,
+      "step": 468
+    },
+    {
+      "epoch": 0.58625,
+      "grad_norm": 2.5906476974487305,
+      "learning_rate": 8.275000000000001e-06,
+      "loss": 3.0516,
+      "step": 469
+    },
+    {
+      "epoch": 0.5875,
+      "grad_norm": 2.569822072982788,
+      "learning_rate": 8.25e-06,
+      "loss": 2.2488,
+      "step": 470
+    },
+    {
+      "epoch": 0.58875,
+      "grad_norm": 1.5829050540924072,
+      "learning_rate": 8.225e-06,
+      "loss": 2.4019,
+      "step": 471
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.332492828369141,
+      "learning_rate": 8.2e-06,
+      "loss": 3.1786,
+      "step": 472
+    },
+    {
+      "epoch": 0.59125,
+      "grad_norm": 2.3091604709625244,
+      "learning_rate": 8.175e-06,
+      "loss": 2.3487,
+      "step": 473
+    },
+    {
+      "epoch": 0.5925,
+      "grad_norm": 1.708343505859375,
+      "learning_rate": 8.15e-06,
+      "loss": 2.5667,
+      "step": 474
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 1.4315516948699951,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 2.4841,
+      "step": 475
+    },
+    {
+      "epoch": 0.595,
+      "grad_norm": 1.4762920141220093,
+      "learning_rate": 8.1e-06,
+      "loss": 2.5809,
+      "step": 476
+    },
+    {
+      "epoch": 0.59625,
+      "grad_norm": 1.6080222129821777,
+      "learning_rate": 8.075000000000001e-06,
+      "loss": 2.6133,
+      "step": 477
+    },
+    {
+      "epoch": 0.5975,
+      "grad_norm": 1.0008354187011719,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 2.3204,
+      "step": 478
+    },
+    {
+      "epoch": 0.59875,
+      "grad_norm": 1.433173656463623,
+      "learning_rate": 8.025e-06,
+      "loss": 2.6281,
+      "step": 479
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1544861793518066,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.7092,
+      "step": 480
+    },
+    {
+      "epoch": 0.60125,
+      "grad_norm": 1.2630051374435425,
+      "learning_rate": 7.975e-06,
+      "loss": 2.2522,
+      "step": 481
+    },
+    {
+      "epoch": 0.6025,
+      "grad_norm": 2.1408400535583496,
+      "learning_rate": 7.950000000000002e-06,
+      "loss": 3.1307,
+      "step": 482
+    },
+    {
+      "epoch": 0.60375,
+      "grad_norm": 1.695609211921692,
+      "learning_rate": 7.925000000000001e-06,
+      "loss": 2.4184,
+      "step": 483
+    },
+    {
+      "epoch": 0.605,
+      "grad_norm": 1.8924223184585571,
+      "learning_rate": 7.9e-06,
+      "loss": 3.2703,
+      "step": 484
+    },
+    {
+      "epoch": 0.60625,
+      "grad_norm": 4.110727787017822,
+      "learning_rate": 7.875e-06,
+      "loss": 3.0813,
+      "step": 485
+    },
+    {
+      "epoch": 0.6075,
+      "grad_norm": 1.2760142087936401,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 2.7032,
+      "step": 486
+    },
+    {
+      "epoch": 0.60875,
+      "grad_norm": 1.2902387380599976,
+      "learning_rate": 7.825e-06,
+      "loss": 2.7254,
+      "step": 487
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.366107940673828,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.8782,
+      "step": 488
+    },
+    {
+      "epoch": 0.61125,
+      "grad_norm": 2.737114906311035,
+      "learning_rate": 7.775000000000001e-06,
+      "loss": 3.5096,
+      "step": 489
+    },
+    {
+      "epoch": 0.6125,
+      "grad_norm": 4.056971073150635,
+      "learning_rate": 7.75e-06,
+      "loss": 3.4208,
+      "step": 490
+    },
+    {
+      "epoch": 0.61375,
+      "grad_norm": 1.18091881275177,
+      "learning_rate": 7.725e-06,
+      "loss": 2.6447,
+      "step": 491
+    },
+    {
+      "epoch": 0.615,
+      "grad_norm": 5.914666652679443,
+      "learning_rate": 7.7e-06,
+      "loss": 4.1583,
+      "step": 492
+    },
+    {
+      "epoch": 0.61625,
+      "grad_norm": 1.383678913116455,
+      "learning_rate": 7.675e-06,
+      "loss": 2.685,
+      "step": 493
+    },
+    {
+      "epoch": 0.6175,
+      "grad_norm": 1.5389842987060547,
+      "learning_rate": 7.650000000000001e-06,
+      "loss": 2.9398,
+      "step": 494
+    },
+    {
+      "epoch": 0.61875,
+      "grad_norm": 1.1608572006225586,
+      "learning_rate": 7.625e-06,
+      "loss": 2.8138,
+      "step": 495
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.245866298675537,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.2935,
+      "step": 496
+    },
+    {
+      "epoch": 0.62125,
+      "grad_norm": 1.0834670066833496,
+      "learning_rate": 7.575e-06,
+      "loss": 2.3084,
+      "step": 497
+    },
+    {
+      "epoch": 0.6225,
+      "grad_norm": 1.079750657081604,
+      "learning_rate": 7.5500000000000006e-06,
+      "loss": 1.861,
+      "step": 498
+    },
+    {
+      "epoch": 0.62375,
+      "grad_norm": 1.2369153499603271,
+      "learning_rate": 7.525e-06,
+      "loss": 2.3253,
+      "step": 499
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.6367318630218506,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.1319,
+      "step": 500
+    },
+    {
+      "epoch": 0.62625,
+      "grad_norm": 1.771266222000122,
+      "learning_rate": 7.475000000000001e-06,
+      "loss": 2.6704,
+      "step": 501
+    },
+    {
+      "epoch": 0.6275,
+      "grad_norm": 2.236402988433838,
+      "learning_rate": 7.450000000000001e-06,
+      "loss": 3.2022,
+      "step": 502
+    },
+    {
+      "epoch": 0.62875,
+      "grad_norm": 1.600723385810852,
+      "learning_rate": 7.425000000000001e-06,
+      "loss": 2.4008,
+      "step": 503
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.818192720413208,
+      "learning_rate": 7.4e-06,
+      "loss": 3.191,
+      "step": 504
+    },
+    {
+      "epoch": 0.63125,
+      "grad_norm": 2.4682326316833496,
+      "learning_rate": 7.375000000000001e-06,
+      "loss": 2.9254,
+      "step": 505
+    },
+    {
+      "epoch": 0.6325,
+      "grad_norm": 1.8725566864013672,
+      "learning_rate": 7.350000000000001e-06,
+      "loss": 2.6224,
+      "step": 506
+    },
+    {
+      "epoch": 0.63375,
+      "grad_norm": 1.5210118293762207,
+      "learning_rate": 7.325000000000001e-06,
+      "loss": 2.2774,
+      "step": 507
+    },
+    {
+      "epoch": 0.635,
+      "grad_norm": 1.918311357498169,
+      "learning_rate": 7.3e-06,
+      "loss": 2.6411,
+      "step": 508
+    },
+    {
+      "epoch": 0.63625,
+      "grad_norm": 0.7823852896690369,
+      "learning_rate": 7.275000000000001e-06,
+      "loss": 2.2566,
+      "step": 509
+    },
+    {
+      "epoch": 0.6375,
+      "grad_norm": 1.1743433475494385,
+      "learning_rate": 7.25e-06,
+      "loss": 2.7283,
+      "step": 510
+    },
+    {
+      "epoch": 0.63875,
+      "grad_norm": 4.318971633911133,
+      "learning_rate": 7.225000000000001e-06,
+      "loss": 3.1266,
+      "step": 511
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3978825807571411,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.3688,
+      "step": 512
+    },
+    {
+      "epoch": 0.64125,
+      "grad_norm": 1.2514866590499878,
+      "learning_rate": 7.175000000000001e-06,
+      "loss": 1.8098,
+      "step": 513
+    },
+    {
+      "epoch": 0.6425,
+      "grad_norm": 2.5508549213409424,
+      "learning_rate": 7.15e-06,
+      "loss": 2.7809,
+      "step": 514
+    },
+    {
+      "epoch": 0.64375,
+      "grad_norm": 1.093632698059082,
+      "learning_rate": 7.125e-06,
+      "loss": 2.6663,
+      "step": 515
+    },
+    {
+      "epoch": 0.645,
+      "grad_norm": 1.880379557609558,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.539,
+      "step": 516
+    },
+    {
+      "epoch": 0.64625,
+      "grad_norm": 3.1032094955444336,
+      "learning_rate": 7.075000000000001e-06,
+      "loss": 2.1864,
+      "step": 517
+    },
+    {
+      "epoch": 0.6475,
+      "grad_norm": 1.828586459159851,
+      "learning_rate": 7.05e-06,
+      "loss": 2.7087,
+      "step": 518
+    },
+    {
+      "epoch": 0.64875,
+      "grad_norm": 5.738358497619629,
+      "learning_rate": 7.0250000000000005e-06,
+      "loss": 3.8585,
+      "step": 519
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2483720779418945,
+      "learning_rate": 7e-06,
+      "loss": 2.424,
+      "step": 520
+    },
+    {
+      "epoch": 0.65125,
+      "grad_norm": 1.6957286596298218,
+      "learning_rate": 6.975000000000001e-06,
+      "loss": 2.053,
+      "step": 521
+    },
+    {
+      "epoch": 0.6525,
+      "grad_norm": 1.2034648656845093,
+      "learning_rate": 6.95e-06,
+      "loss": 1.6862,
+      "step": 522
+    },
+    {
+      "epoch": 0.65375,
+      "grad_norm": 2.2950422763824463,
+      "learning_rate": 6.925000000000001e-06,
+      "loss": 2.6543,
+      "step": 523
+    },
+    {
+      "epoch": 0.655,
+      "grad_norm": 3.466404438018799,
+      "learning_rate": 6.9e-06,
+      "loss": 3.5708,
+      "step": 524
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 1.826366662979126,
+      "learning_rate": 6.875e-06,
+      "loss": 2.5286,
+      "step": 525
+    },
+    {
+      "epoch": 0.6575,
+      "grad_norm": 1.3622981309890747,
+      "learning_rate": 6.850000000000001e-06,
+      "loss": 2.3816,
+      "step": 526
+    },
+    {
+      "epoch": 0.65875,
+      "grad_norm": 1.7400751113891602,
+      "learning_rate": 6.825000000000001e-06,
+      "loss": 2.1949,
+      "step": 527
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.261826753616333,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.1731,
+      "step": 528
+    },
+    {
+      "epoch": 0.66125,
+      "grad_norm": 1.299221158027649,
+      "learning_rate": 6.775e-06,
+      "loss": 2.664,
+      "step": 529
+    },
+    {
+      "epoch": 0.6625,
+      "grad_norm": 5.387721538543701,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 3.4304,
+      "step": 530
+    },
+    {
+      "epoch": 0.66375,
+      "grad_norm": 2.0014383792877197,
+      "learning_rate": 6.725000000000001e-06,
+      "loss": 2.4377,
+      "step": 531
+    },
+    {
+      "epoch": 0.665,
+      "grad_norm": 1.8447600603103638,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.0142,
+      "step": 532
+    },
+    {
+      "epoch": 0.66625,
+      "grad_norm": 2.1541025638580322,
+      "learning_rate": 6.6750000000000005e-06,
+      "loss": 2.3221,
+      "step": 533
+    },
+    {
+      "epoch": 0.6675,
+      "grad_norm": 2.365069627761841,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 2.8227,
+      "step": 534
+    },
+    {
+      "epoch": 0.66875,
+      "grad_norm": 2.2634615898132324,
+      "learning_rate": 6.625e-06,
+      "loss": 2.7516,
+      "step": 535
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8768017292022705,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.129,
+      "step": 536
+    },
+    {
+      "epoch": 0.67125,
+      "grad_norm": 2.3933587074279785,
+      "learning_rate": 6.5750000000000006e-06,
+      "loss": 2.457,
+      "step": 537
+    },
+    {
+      "epoch": 0.6725,
+      "grad_norm": 1.068185806274414,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 2.2805,
+      "step": 538
+    },
+    {
+      "epoch": 0.67375,
+      "grad_norm": 1.4682180881500244,
+      "learning_rate": 6.525e-06,
+      "loss": 2.4401,
+      "step": 539
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 2.4375839233398438,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.0402,
+      "step": 540
+    },
+    {
+      "epoch": 0.67625,
+      "grad_norm": 1.7654716968536377,
+      "learning_rate": 6.475e-06,
+      "loss": 2.4122,
+      "step": 541
+    },
+    {
+      "epoch": 0.6775,
+      "grad_norm": 2.0273666381835938,
+      "learning_rate": 6.450000000000001e-06,
+      "loss": 2.7685,
+      "step": 542
+    },
+    {
+      "epoch": 0.67875,
+      "grad_norm": 2.323498487472534,
+      "learning_rate": 6.425e-06,
+      "loss": 2.735,
+      "step": 543
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4310922622680664,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.0557,
+      "step": 544
+    },
+    {
+      "epoch": 0.68125,
+      "grad_norm": 2.29921817779541,
+      "learning_rate": 6.375e-06,
+      "loss": 2.6454,
+      "step": 545
+    },
+    {
+      "epoch": 0.6825,
+      "grad_norm": 5.133302211761475,
+      "learning_rate": 6.35e-06,
+      "loss": 3.3667,
+      "step": 546
+    },
+    {
+      "epoch": 0.68375,
+      "grad_norm": 1.5254071950912476,
+      "learning_rate": 6.3250000000000004e-06,
+      "loss": 2.1213,
+      "step": 547
+    },
+    {
+      "epoch": 0.685,
+      "grad_norm": 2.4520602226257324,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.9527,
+      "step": 548
+    },
+    {
+      "epoch": 0.68625,
+      "grad_norm": 3.5153932571411133,
+      "learning_rate": 6.275e-06,
+      "loss": 2.6369,
+      "step": 549
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 1.860868215560913,
+      "learning_rate": 6.25e-06,
+      "loss": 3.4027,
+      "step": 550
+    },
+    {
+      "epoch": 0.68875,
+      "grad_norm": 2.164926290512085,
+      "learning_rate": 6.225000000000001e-06,
+      "loss": 2.5587,
+      "step": 551
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0053963661193848,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.7041,
+      "step": 552
+    },
+    {
+      "epoch": 0.69125,
+      "grad_norm": 1.467716932296753,
+      "learning_rate": 6.175000000000001e-06,
+      "loss": 2.4994,
+      "step": 553
+    },
+    {
+      "epoch": 0.6925,
+      "grad_norm": 1.1203712224960327,
+      "learning_rate": 6.15e-06,
+      "loss": 2.8008,
+      "step": 554
+    },
+    {
+      "epoch": 0.69375,
+      "grad_norm": 1.4410206079483032,
+      "learning_rate": 6.125000000000001e-06,
+      "loss": 2.6894,
+      "step": 555
+    },
+    {
+      "epoch": 0.695,
+      "grad_norm": 2.3052964210510254,
+      "learning_rate": 6.1e-06,
+      "loss": 2.7815,
+      "step": 556
+    },
+    {
+      "epoch": 0.69625,
+      "grad_norm": 1.6395195722579956,
+      "learning_rate": 6.075000000000001e-06,
+      "loss": 2.3493,
+      "step": 557
+    },
+    {
+      "epoch": 0.6975,
+      "grad_norm": 0.6925020217895508,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 2.375,
+      "step": 558
+    },
+    {
+      "epoch": 0.69875,
+      "grad_norm": 2.1917574405670166,
+      "learning_rate": 6.025000000000001e-06,
+      "loss": 3.4288,
+      "step": 559
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4762030839920044,
+      "learning_rate": 6e-06,
+      "loss": 2.8223,
+      "step": 560
+    },
+    {
+      "epoch": 0.70125,
+      "grad_norm": 2.2189676761627197,
+      "learning_rate": 5.975e-06,
+      "loss": 3.0123,
+      "step": 561
+    },
+    {
+      "epoch": 0.7025,
+      "grad_norm": 2.046457052230835,
+      "learning_rate": 5.950000000000001e-06,
+      "loss": 2.1167,
+      "step": 562
+    },
+    {
+      "epoch": 0.70375,
+      "grad_norm": 2.0804660320281982,
+      "learning_rate": 5.925000000000001e-06,
+      "loss": 2.4246,
+      "step": 563
+    },
+    {
+      "epoch": 0.705,
+      "grad_norm": 2.419306993484497,
+      "learning_rate": 5.9e-06,
+      "loss": 2.2055,
+      "step": 564
+    },
+    {
+      "epoch": 0.70625,
+      "grad_norm": 1.8702435493469238,
+      "learning_rate": 5.8750000000000005e-06,
+      "loss": 2.6608,
+      "step": 565
+    },
+    {
+      "epoch": 0.7075,
+      "grad_norm": 2.246138572692871,
+      "learning_rate": 5.85e-06,
+      "loss": 2.7353,
+      "step": 566
+    },
+    {
+      "epoch": 0.70875,
+      "grad_norm": 1.2351369857788086,
+      "learning_rate": 5.825000000000001e-06,
+      "loss": 2.0033,
+      "step": 567
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4821653366088867,
+      "learning_rate": 5.8e-06,
+      "loss": 2.9512,
+      "step": 568
+    },
+    {
+      "epoch": 0.71125,
+      "grad_norm": 1.1718158721923828,
+      "learning_rate": 5.775000000000001e-06,
+      "loss": 2.5153,
+      "step": 569
+    },
+    {
+      "epoch": 0.7125,
+      "grad_norm": 1.8521480560302734,
+      "learning_rate": 5.75e-06,
+      "loss": 3.0076,
+      "step": 570
+    },
+    {
+      "epoch": 0.71375,
+      "grad_norm": 1.7520173788070679,
+      "learning_rate": 5.725e-06,
+      "loss": 2.4602,
+      "step": 571
+    },
+    {
+      "epoch": 0.715,
+      "grad_norm": 3.2388157844543457,
+      "learning_rate": 5.7e-06,
+      "loss": 3.1706,
+      "step": 572
+    },
+    {
+      "epoch": 0.71625,
+      "grad_norm": 1.5112570524215698,
+      "learning_rate": 5.675000000000001e-06,
+      "loss": 2.5468,
+      "step": 573
+    },
+    {
+      "epoch": 0.7175,
+      "grad_norm": 2.0039563179016113,
+      "learning_rate": 5.65e-06,
+      "loss": 2.98,
+      "step": 574
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 2.5694010257720947,
+      "learning_rate": 5.625e-06,
+      "loss": 2.454,
+      "step": 575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8285837173461914,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.4644,
+      "step": 576
+    },
+    {
+      "epoch": 0.72125,
+      "grad_norm": 1.8699629306793213,
+      "learning_rate": 5.575000000000001e-06,
+      "loss": 2.0484,
+      "step": 577
+    },
+    {
+      "epoch": 0.7225,
+      "grad_norm": 1.4237912893295288,
+      "learning_rate": 5.550000000000001e-06,
+      "loss": 2.5569,
+      "step": 578
+    },
+    {
+      "epoch": 0.72375,
+      "grad_norm": 1.812650442123413,
+      "learning_rate": 5.5250000000000005e-06,
+      "loss": 2.8348,
+      "step": 579
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 4.9448137283325195,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 4.1679,
+      "step": 580
+    },
+    {
+      "epoch": 0.72625,
+      "grad_norm": 3.0024564266204834,
+      "learning_rate": 5.475e-06,
+      "loss": 3.1915,
+      "step": 581
+    },
+    {
+      "epoch": 0.7275,
+      "grad_norm": 1.2082493305206299,
+      "learning_rate": 5.450000000000001e-06,
+      "loss": 1.7674,
+      "step": 582
+    },
+    {
+      "epoch": 0.72875,
+      "grad_norm": 2.0212562084198,
+      "learning_rate": 5.4250000000000006e-06,
+      "loss": 2.8302,
+      "step": 583
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7197890281677246,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.0274,
+      "step": 584
+    },
+    {
+      "epoch": 0.73125,
+      "grad_norm": 1.3871604204177856,
+      "learning_rate": 5.375e-06,
+      "loss": 2.4914,
+      "step": 585
+    },
+    {
+      "epoch": 0.7325,
+      "grad_norm": 1.2983068227767944,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 3.1802,
+      "step": 586
+    },
+    {
+      "epoch": 0.73375,
+      "grad_norm": 1.3344168663024902,
+      "learning_rate": 5.325e-06,
+      "loss": 2.1649,
+      "step": 587
+    },
+    {
+      "epoch": 0.735,
+      "grad_norm": 1.4151086807250977,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.8521,
+      "step": 588
+    },
+    {
+      "epoch": 0.73625,
+      "grad_norm": 2.3366806507110596,
+      "learning_rate": 5.275e-06,
+      "loss": 2.8001,
+      "step": 589
+    },
+    {
+      "epoch": 0.7375,
+      "grad_norm": 3.088120460510254,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 2.4478,
+      "step": 590
+    },
+    {
+      "epoch": 0.73875,
+      "grad_norm": 3.974681854248047,
+      "learning_rate": 5.225e-06,
+      "loss": 2.633,
+      "step": 591
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.696179151535034,
+      "learning_rate": 5.2e-06,
+      "loss": 2.8837,
+      "step": 592
+    },
+    {
+      "epoch": 0.74125,
+      "grad_norm": 2.7638463973999023,
+      "learning_rate": 5.1750000000000004e-06,
+      "loss": 2.4203,
+      "step": 593
+    },
+    {
+      "epoch": 0.7425,
+      "grad_norm": 1.9097126722335815,
+      "learning_rate": 5.150000000000001e-06,
+      "loss": 2.1879,
+      "step": 594
+    },
+    {
+      "epoch": 0.74375,
+      "grad_norm": 1.0455313920974731,
+      "learning_rate": 5.125e-06,
+      "loss": 2.529,
+      "step": 595
+    },
+    {
+      "epoch": 0.745,
+      "grad_norm": 1.6425269842147827,
+      "learning_rate": 5.1e-06,
+      "loss": 3.3192,
+      "step": 596
+    },
+    {
+      "epoch": 0.74625,
+      "grad_norm": 0.8040964007377625,
+      "learning_rate": 5.075e-06,
+      "loss": 2.2812,
+      "step": 597
+    },
+    {
+      "epoch": 0.7475,
+      "grad_norm": 1.8883963823318481,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 3.0112,
+      "step": 598
+    },
+    {
+      "epoch": 0.74875,
+      "grad_norm": 1.580411434173584,
+      "learning_rate": 5.025e-06,
+      "loss": 2.7145,
+      "step": 599
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.2164394855499268,
+      "learning_rate": 5e-06,
+      "loss": 2.0817,
+      "step": 600
+    },
+    {
+      "epoch": 0.75125,
+      "grad_norm": 1.7922728061676025,
+      "learning_rate": 4.975000000000001e-06,
+      "loss": 2.4854,
+      "step": 601
+    },
+    {
+      "epoch": 0.7525,
+      "grad_norm": 1.1481162309646606,
+      "learning_rate": 4.95e-06,
+      "loss": 2.5298,
+      "step": 602
+    },
+    {
+      "epoch": 0.75375,
+      "grad_norm": 1.3136919736862183,
+      "learning_rate": 4.925e-06,
+      "loss": 2.4225,
+      "step": 603
+    },
+    {
+      "epoch": 0.755,
+      "grad_norm": 2.7019455432891846,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.8419,
+      "step": 604
+    },
+    {
+      "epoch": 0.75625,
+      "grad_norm": 1.661961555480957,
+      "learning_rate": 4.875e-06,
+      "loss": 2.097,
+      "step": 605
+    },
+    {
+      "epoch": 0.7575,
+      "grad_norm": 1.692785620689392,
+      "learning_rate": 4.85e-06,
+      "loss": 2.7115,
+      "step": 606
+    },
+    {
+      "epoch": 0.75875,
+      "grad_norm": 3.2370035648345947,
+      "learning_rate": 4.825e-06,
+      "loss": 2.7567,
+      "step": 607
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9687768220901489,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.362,
+      "step": 608
+    },
+    {
+      "epoch": 0.76125,
+      "grad_norm": 1.3951308727264404,
+      "learning_rate": 4.775e-06,
+      "loss": 2.2084,
+      "step": 609
+    },
+    {
+      "epoch": 0.7625,
+      "grad_norm": 2.4143717288970947,
+      "learning_rate": 4.75e-06,
+      "loss": 2.3073,
+      "step": 610
+    },
+    {
+      "epoch": 0.76375,
+      "grad_norm": 1.0696815252304077,
+      "learning_rate": 4.7250000000000005e-06,
+      "loss": 1.8974,
+      "step": 611
+    },
+    {
+      "epoch": 0.765,
+      "grad_norm": 1.8856892585754395,
+      "learning_rate": 4.7e-06,
+      "loss": 2.6322,
+      "step": 612
+    },
+    {
+      "epoch": 0.76625,
+      "grad_norm": 2.168942451477051,
+      "learning_rate": 4.675000000000001e-06,
+      "loss": 2.5815,
+      "step": 613
+    },
+    {
+      "epoch": 0.7675,
+      "grad_norm": 2.5787465572357178,
+      "learning_rate": 4.65e-06,
+      "loss": 3.1303,
+      "step": 614
+    },
+    {
+      "epoch": 0.76875,
+      "grad_norm": 1.4108450412750244,
+      "learning_rate": 4.625000000000001e-06,
+      "loss": 2.44,
+      "step": 615
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7232167720794678,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.4796,
+      "step": 616
+    },
+    {
+      "epoch": 0.77125,
+      "grad_norm": 1.3400771617889404,
+      "learning_rate": 4.575e-06,
+      "loss": 2.5869,
+      "step": 617
+    },
+    {
+      "epoch": 0.7725,
+      "grad_norm": 1.1327736377716064,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 1.856,
+      "step": 618
+    },
+    {
+      "epoch": 0.77375,
+      "grad_norm": 1.3946475982666016,
+      "learning_rate": 4.525000000000001e-06,
+      "loss": 2.4039,
+      "step": 619
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 2.2213644981384277,
+      "learning_rate": 4.5e-06,
+      "loss": 2.6722,
+      "step": 620
+    },
+    {
+      "epoch": 0.77625,
+      "grad_norm": 3.5649960041046143,
+      "learning_rate": 4.475e-06,
+      "loss": 2.5282,
+      "step": 621
+    },
+    {
+      "epoch": 0.7775,
+      "grad_norm": 1.5441259145736694,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 2.6255,
+      "step": 622
+    },
+    {
+      "epoch": 0.77875,
+      "grad_norm": 1.637509822845459,
+      "learning_rate": 4.425e-06,
+      "loss": 2.3554,
+      "step": 623
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0805779695510864,
+      "learning_rate": 4.4e-06,
+      "loss": 2.543,
+      "step": 624
+    },
+    {
+      "epoch": 0.78125,
+      "grad_norm": 2.173614740371704,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 2.2831,
+      "step": 625
+    },
+    {
+      "epoch": 0.7825,
+      "grad_norm": 1.3933883905410767,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 2.3316,
+      "step": 626
+    },
+    {
+      "epoch": 0.78375,
+      "grad_norm": 0.9714035987854004,
+      "learning_rate": 4.325e-06,
+      "loss": 2.344,
+      "step": 627
+    },
+    {
+      "epoch": 0.785,
+      "grad_norm": 1.6793344020843506,
+      "learning_rate": 4.3e-06,
+      "loss": 2.5151,
+      "step": 628
+    },
+    {
+      "epoch": 0.78625,
+      "grad_norm": 1.507856011390686,
+      "learning_rate": 4.2750000000000006e-06,
+      "loss": 3.1518,
+      "step": 629
+    },
+    {
+      "epoch": 0.7875,
+      "grad_norm": 3.8492484092712402,
+      "learning_rate": 4.25e-06,
+      "loss": 3.2012,
+      "step": 630
+    },
+    {
+      "epoch": 0.78875,
+      "grad_norm": 2.8230409622192383,
+      "learning_rate": 4.225e-06,
+      "loss": 3.0812,
+      "step": 631
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.977073311805725,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.4044,
+      "step": 632
+    },
+    {
+      "epoch": 0.79125,
+      "grad_norm": 2.227055788040161,
+      "learning_rate": 4.175e-06,
+      "loss": 2.9186,
+      "step": 633
+    },
+    {
+      "epoch": 0.7925,
+      "grad_norm": 3.2962162494659424,
+      "learning_rate": 4.15e-06,
+      "loss": 3.5546,
+      "step": 634
+    },
+    {
+      "epoch": 0.79375,
+      "grad_norm": 1.9414913654327393,
+      "learning_rate": 4.125e-06,
+      "loss": 2.7475,
+      "step": 635
+    },
+    {
+      "epoch": 0.795,
+      "grad_norm": 1.6483408212661743,
+      "learning_rate": 4.1e-06,
+      "loss": 3.0241,
+      "step": 636
+    },
+    {
+      "epoch": 0.79625,
+      "grad_norm": 1.1849037408828735,
+      "learning_rate": 4.075e-06,
+      "loss": 2.6205,
+      "step": 637
+    },
+    {
+      "epoch": 0.7975,
+      "grad_norm": 1.7737408876419067,
+      "learning_rate": 4.05e-06,
+      "loss": 2.5812,
+      "step": 638
+    },
+    {
+      "epoch": 0.79875,
+      "grad_norm": 1.0205885171890259,
+      "learning_rate": 4.0250000000000004e-06,
+      "loss": 2.5972,
+      "step": 639
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9625855088233948,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.2021,
+      "step": 640
+    },
+    {
+      "epoch": 0.80125,
+      "grad_norm": 1.3905662298202515,
+      "learning_rate": 3.975000000000001e-06,
+      "loss": 2.1898,
+      "step": 641
+    },
+    {
+      "epoch": 0.8025,
+      "grad_norm": 2.1168220043182373,
+      "learning_rate": 3.95e-06,
+      "loss": 2.866,
+      "step": 642
+    },
+    {
+      "epoch": 0.80375,
+      "grad_norm": 3.0257530212402344,
+      "learning_rate": 3.9250000000000005e-06,
+      "loss": 2.8517,
+      "step": 643
+    },
+    {
+      "epoch": 0.805,
+      "grad_norm": 2.563762903213501,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.3336,
+      "step": 644
+    },
+    {
+      "epoch": 0.80625,
+      "grad_norm": 3.3193588256835938,
+      "learning_rate": 3.875e-06,
+      "loss": 3.1035,
+      "step": 645
+    },
+    {
+      "epoch": 0.8075,
+      "grad_norm": 2.2538487911224365,
+      "learning_rate": 3.85e-06,
+      "loss": 2.8489,
+      "step": 646
+    },
+    {
+      "epoch": 0.80875,
+      "grad_norm": 1.6400374174118042,
+      "learning_rate": 3.825000000000001e-06,
+      "loss": 2.7454,
+      "step": 647
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.152010202407837,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.5955,
+      "step": 648
+    },
+    {
+      "epoch": 0.81125,
+      "grad_norm": 1.100693941116333,
+      "learning_rate": 3.7750000000000003e-06,
+      "loss": 1.9502,
+      "step": 649
+    },
+    {
+      "epoch": 0.8125,
+      "grad_norm": 1.8576849699020386,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.7005,
+      "step": 650
+    },
+    {
+      "epoch": 0.81375,
+      "grad_norm": 2.48925518989563,
+      "learning_rate": 3.7250000000000003e-06,
+      "loss": 2.7473,
+      "step": 651
+    },
+    {
+      "epoch": 0.815,
+      "grad_norm": 1.4268637895584106,
+      "learning_rate": 3.7e-06,
+      "loss": 2.4413,
+      "step": 652
+    },
+    {
+      "epoch": 0.81625,
+      "grad_norm": 0.8600934743881226,
+      "learning_rate": 3.6750000000000004e-06,
+      "loss": 2.7305,
+      "step": 653
+    },
+    {
+      "epoch": 0.8175,
+      "grad_norm": 1.7161887884140015,
+      "learning_rate": 3.65e-06,
+      "loss": 2.0339,
+      "step": 654
+    },
+    {
+      "epoch": 0.81875,
+      "grad_norm": 2.643136739730835,
+      "learning_rate": 3.625e-06,
+      "loss": 2.7862,
+      "step": 655
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.1652709245681763,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.3251,
+      "step": 656
+    },
+    {
+      "epoch": 0.82125,
+      "grad_norm": 1.3739567995071411,
+      "learning_rate": 3.575e-06,
+      "loss": 2.5092,
+      "step": 657
+    },
+    {
+      "epoch": 0.8225,
+      "grad_norm": 1.4893128871917725,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 2.2714,
+      "step": 658
+    },
+    {
+      "epoch": 0.82375,
+      "grad_norm": 0.9546101093292236,
+      "learning_rate": 3.525e-06,
+      "loss": 2.106,
+      "step": 659
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 1.596313238143921,
+      "learning_rate": 3.5e-06,
+      "loss": 2.9919,
+      "step": 660
+    },
+    {
+      "epoch": 0.82625,
+      "grad_norm": 1.6426167488098145,
+      "learning_rate": 3.475e-06,
+      "loss": 2.2044,
+      "step": 661
+    },
+    {
+      "epoch": 0.8275,
+      "grad_norm": 1.0089974403381348,
+      "learning_rate": 3.45e-06,
+      "loss": 2.5872,
+      "step": 662
+    },
+    {
+      "epoch": 0.82875,
+      "grad_norm": 1.9362369775772095,
+      "learning_rate": 3.4250000000000007e-06,
+      "loss": 2.5531,
+      "step": 663
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6755287051200867,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.9512,
+      "step": 664
+    },
+    {
+      "epoch": 0.83125,
+      "grad_norm": 1.8554205894470215,
+      "learning_rate": 3.3750000000000003e-06,
+      "loss": 2.5813,
+      "step": 665
+    },
+    {
+      "epoch": 0.8325,
+      "grad_norm": 2.031695604324341,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 2.6638,
+      "step": 666
+    },
+    {
+      "epoch": 0.83375,
+      "grad_norm": 1.817267656326294,
+      "learning_rate": 3.3250000000000004e-06,
+      "loss": 2.7071,
+      "step": 667
+    },
+    {
+      "epoch": 0.835,
+      "grad_norm": 1.8573476076126099,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.9174,
+      "step": 668
+    },
+    {
+      "epoch": 0.83625,
+      "grad_norm": 1.9300901889801025,
+      "learning_rate": 3.2750000000000004e-06,
+      "loss": 3.0282,
+      "step": 669
+    },
+    {
+      "epoch": 0.8375,
+      "grad_norm": 0.8869524598121643,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 2.4725,
+      "step": 670
+    },
+    {
+      "epoch": 0.83875,
+      "grad_norm": 1.1387746334075928,
+      "learning_rate": 3.2250000000000005e-06,
+      "loss": 2.3742,
+      "step": 671
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.481876850128174,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.567,
+      "step": 672
+    },
+    {
+      "epoch": 0.84125,
+      "grad_norm": 1.3130849599838257,
+      "learning_rate": 3.175e-06,
+      "loss": 2.1865,
+      "step": 673
+    },
+    {
+      "epoch": 0.8425,
+      "grad_norm": 1.5404133796691895,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 2.6415,
+      "step": 674
+    },
+    {
+      "epoch": 0.84375,
+      "grad_norm": 2.30031418800354,
+      "learning_rate": 3.125e-06,
+      "loss": 2.6346,
+      "step": 675
+    },
+    {
+      "epoch": 0.845,
+      "grad_norm": 1.8086575269699097,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.1789,
+      "step": 676
+    },
+    {
+      "epoch": 0.84625,
+      "grad_norm": 2.2194817066192627,
+      "learning_rate": 3.075e-06,
+      "loss": 2.048,
+      "step": 677
+    },
+    {
+      "epoch": 0.8475,
+      "grad_norm": 1.6533406972885132,
+      "learning_rate": 3.05e-06,
+      "loss": 2.3017,
+      "step": 678
+    },
+    {
+      "epoch": 0.84875,
+      "grad_norm": 4.319630146026611,
+      "learning_rate": 3.0250000000000003e-06,
+      "loss": 3.3279,
+      "step": 679
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.66692316532135,
+      "learning_rate": 3e-06,
+      "loss": 2.6651,
+      "step": 680
+    },
+    {
+      "epoch": 0.85125,
+      "grad_norm": 0.786642849445343,
+      "learning_rate": 2.9750000000000003e-06,
+      "loss": 1.92,
+      "step": 681
+    },
+    {
+      "epoch": 0.8525,
+      "grad_norm": 1.6372087001800537,
+      "learning_rate": 2.95e-06,
+      "loss": 2.394,
+      "step": 682
+    },
+    {
+      "epoch": 0.85375,
+      "grad_norm": 1.3699547052383423,
+      "learning_rate": 2.925e-06,
+      "loss": 2.5697,
+      "step": 683
+    },
+    {
+      "epoch": 0.855,
+      "grad_norm": 2.5320935249328613,
+      "learning_rate": 2.9e-06,
+      "loss": 2.8745,
+      "step": 684
+    },
+    {
+      "epoch": 0.85625,
+      "grad_norm": 2.3555850982666016,
+      "learning_rate": 2.875e-06,
+      "loss": 2.7129,
+      "step": 685
+    },
+    {
+      "epoch": 0.8575,
+      "grad_norm": 2.2113137245178223,
+      "learning_rate": 2.85e-06,
+      "loss": 3.0649,
+      "step": 686
+    },
+    {
+      "epoch": 0.85875,
+      "grad_norm": 2.225691318511963,
+      "learning_rate": 2.825e-06,
+      "loss": 2.1795,
+      "step": 687
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7828199863433838,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.5819,
+      "step": 688
+    },
+    {
+      "epoch": 0.86125,
+      "grad_norm": 1.4047306776046753,
+      "learning_rate": 2.7750000000000005e-06,
+      "loss": 2.4888,
+      "step": 689
+    },
+    {
+      "epoch": 0.8625,
+      "grad_norm": 3.13736891746521,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 3.2848,
+      "step": 690
+    },
+    {
+      "epoch": 0.86375,
+      "grad_norm": 1.6474286317825317,
+      "learning_rate": 2.7250000000000006e-06,
+      "loss": 2.9487,
+      "step": 691
+    },
+    {
+      "epoch": 0.865,
+      "grad_norm": 1.1393457651138306,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.6245,
+      "step": 692
+    },
+    {
+      "epoch": 0.86625,
+      "grad_norm": 2.4159979820251465,
+      "learning_rate": 2.6750000000000002e-06,
+      "loss": 2.5928,
+      "step": 693
+    },
+    {
+      "epoch": 0.8675,
+      "grad_norm": 2.7899107933044434,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 3.0244,
+      "step": 694
+    },
+    {
+      "epoch": 0.86875,
+      "grad_norm": 1.7800438404083252,
+      "learning_rate": 2.6250000000000003e-06,
+      "loss": 1.7862,
+      "step": 695
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.111862897872925,
+      "learning_rate": 2.6e-06,
+      "loss": 2.8517,
+      "step": 696
+    },
+    {
+      "epoch": 0.87125,
+      "grad_norm": 2.0922229290008545,
+      "learning_rate": 2.5750000000000003e-06,
+      "loss": 2.2728,
+      "step": 697
+    },
+    {
+      "epoch": 0.8725,
+      "grad_norm": 1.8772225379943848,
+      "learning_rate": 2.55e-06,
+      "loss": 2.3808,
+      "step": 698
+    },
+    {
+      "epoch": 0.87375,
+      "grad_norm": 1.4865831136703491,
+      "learning_rate": 2.5250000000000004e-06,
+      "loss": 2.814,
+      "step": 699
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 2.728365182876587,
+      "learning_rate": 2.5e-06,
+      "loss": 3.2643,
+      "step": 700
+    },
+    {
+      "epoch": 0.87625,
+      "grad_norm": 2.335869789123535,
+      "learning_rate": 2.475e-06,
+      "loss": 2.5471,
+      "step": 701
+    },
+    {
+      "epoch": 0.8775,
+      "grad_norm": 2.8404269218444824,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 3.1107,
+      "step": 702
+    },
+    {
+      "epoch": 0.87875,
+      "grad_norm": 2.249152898788452,
+      "learning_rate": 2.425e-06,
+      "loss": 2.2618,
+      "step": 703
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.662062168121338,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.6716,
+      "step": 704
+    },
+    {
+      "epoch": 0.88125,
+      "grad_norm": 0.9941419363021851,
+      "learning_rate": 2.375e-06,
+      "loss": 2.3987,
+      "step": 705
+    },
+    {
+      "epoch": 0.8825,
+      "grad_norm": 0.5474485754966736,
+      "learning_rate": 2.35e-06,
+      "loss": 2.429,
+      "step": 706
+    },
+    {
+      "epoch": 0.88375,
+      "grad_norm": 1.473949909210205,
+      "learning_rate": 2.325e-06,
+      "loss": 2.5365,
+      "step": 707
+    },
+    {
+      "epoch": 0.885,
+      "grad_norm": 1.6973087787628174,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.1838,
+      "step": 708
+    },
+    {
+      "epoch": 0.88625,
+      "grad_norm": 1.3658370971679688,
+      "learning_rate": 2.2750000000000002e-06,
+      "loss": 2.7352,
+      "step": 709
+    },
+    {
+      "epoch": 0.8875,
+      "grad_norm": 1.7037274837493896,
+      "learning_rate": 2.25e-06,
+      "loss": 2.4561,
+      "step": 710
+    },
+    {
+      "epoch": 0.88875,
+      "grad_norm": 1.5743016004562378,
+      "learning_rate": 2.2250000000000003e-06,
+      "loss": 2.4667,
+      "step": 711
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2770535945892334,
+      "learning_rate": 2.2e-06,
+      "loss": 2.315,
+      "step": 712
+    },
+    {
+      "epoch": 0.89125,
+      "grad_norm": 1.4078290462493896,
+      "learning_rate": 2.1750000000000004e-06,
+      "loss": 1.8871,
+      "step": 713
+    },
+    {
+      "epoch": 0.8925,
+      "grad_norm": 4.485980987548828,
+      "learning_rate": 2.15e-06,
+      "loss": 3.6076,
+      "step": 714
+    },
+    {
+      "epoch": 0.89375,
+      "grad_norm": 1.9531930685043335,
+      "learning_rate": 2.125e-06,
+      "loss": 2.9939,
+      "step": 715
+    },
+    {
+      "epoch": 0.895,
+      "grad_norm": 1.6232836246490479,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.7136,
+      "step": 716
+    },
+    {
+      "epoch": 0.89625,
+      "grad_norm": 1.2092350721359253,
+      "learning_rate": 2.075e-06,
+      "loss": 2.693,
+      "step": 717
+    },
+    {
+      "epoch": 0.8975,
+      "grad_norm": 3.6011886596679688,
+      "learning_rate": 2.05e-06,
+      "loss": 2.6018,
+      "step": 718
+    },
+    {
+      "epoch": 0.89875,
+      "grad_norm": 1.6210492849349976,
+      "learning_rate": 2.025e-06,
+      "loss": 2.3993,
+      "step": 719
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.093855142593384,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.8579,
+      "step": 720
+    },
+    {
+      "epoch": 0.90125,
+      "grad_norm": 1.624720811843872,
+      "learning_rate": 1.975e-06,
+      "loss": 2.5612,
+      "step": 721
+    },
+    {
+      "epoch": 0.9025,
+      "grad_norm": 2.27695631980896,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 1.6602,
+      "step": 722
+    },
+    {
+      "epoch": 0.90375,
+      "grad_norm": 0.9946104884147644,
+      "learning_rate": 1.925e-06,
+      "loss": 2.2992,
+      "step": 723
+    },
+    {
+      "epoch": 0.905,
+      "grad_norm": 1.5145623683929443,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.9268,
+      "step": 724
+    },
+    {
+      "epoch": 0.90625,
+      "grad_norm": 1.1302818059921265,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 2.3965,
+      "step": 725
+    },
+    {
+      "epoch": 0.9075,
+      "grad_norm": 1.6978087425231934,
+      "learning_rate": 1.85e-06,
+      "loss": 2.4108,
+      "step": 726
+    },
+    {
+      "epoch": 0.90875,
+      "grad_norm": 2.2601537704467773,
+      "learning_rate": 1.825e-06,
+      "loss": 2.3154,
+      "step": 727
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.95534348487854,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.6331,
+      "step": 728
+    },
+    {
+      "epoch": 0.91125,
+      "grad_norm": 1.851906418800354,
+      "learning_rate": 1.7750000000000002e-06,
+      "loss": 2.5358,
+      "step": 729
+    },
+    {
+      "epoch": 0.9125,
+      "grad_norm": 1.785172700881958,
+      "learning_rate": 1.75e-06,
+      "loss": 1.9176,
+      "step": 730
+    },
+    {
+      "epoch": 0.91375,
+      "grad_norm": 1.4503031969070435,
+      "learning_rate": 1.725e-06,
+      "loss": 2.1631,
+      "step": 731
+    },
+    {
+      "epoch": 0.915,
+      "grad_norm": 1.4128133058547974,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.4415,
+      "step": 732
+    },
+    {
+      "epoch": 0.91625,
+      "grad_norm": 0.8669462203979492,
+      "learning_rate": 1.6750000000000003e-06,
+      "loss": 2.3275,
+      "step": 733
+    },
+    {
+      "epoch": 0.9175,
+      "grad_norm": 2.6948771476745605,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 2.0348,
+      "step": 734
+    },
+    {
+      "epoch": 0.91875,
+      "grad_norm": 1.6759194135665894,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 2.2684,
+      "step": 735
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6948565244674683,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.416,
+      "step": 736
+    },
+    {
+      "epoch": 0.92125,
+      "grad_norm": 2.1370370388031006,
+      "learning_rate": 1.5750000000000002e-06,
+      "loss": 1.907,
+      "step": 737
+    },
+    {
+      "epoch": 0.9225,
+      "grad_norm": 2.4928674697875977,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 2.2468,
+      "step": 738
+    },
+    {
+      "epoch": 0.92375,
+      "grad_norm": 1.3928344249725342,
+      "learning_rate": 1.525e-06,
+      "loss": 2.9909,
+      "step": 739
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 0.9298745393753052,
+      "learning_rate": 1.5e-06,
+      "loss": 2.2005,
+      "step": 740
+    },
+    {
+      "epoch": 0.92625,
+      "grad_norm": 1.4485491514205933,
+      "learning_rate": 1.475e-06,
+      "loss": 2.1133,
+      "step": 741
+    },
+    {
+      "epoch": 0.9275,
+      "grad_norm": 1.4372122287750244,
+      "learning_rate": 1.45e-06,
+      "loss": 2.923,
+      "step": 742
+    },
+    {
+      "epoch": 0.92875,
+      "grad_norm": 0.9330675005912781,
+      "learning_rate": 1.425e-06,
+      "loss": 2.4085,
+      "step": 743
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9367544651031494,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.8133,
+      "step": 744
+    },
+    {
+      "epoch": 0.93125,
+      "grad_norm": 2.0561740398406982,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 2.6545,
+      "step": 745
+    },
+    {
+      "epoch": 0.9325,
+      "grad_norm": 3.575596332550049,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 2.4606,
+      "step": 746
+    },
+    {
+      "epoch": 0.93375,
+      "grad_norm": 1.240574836730957,
+      "learning_rate": 1.3250000000000002e-06,
+      "loss": 2.7273,
+      "step": 747
+    },
+    {
+      "epoch": 0.935,
+      "grad_norm": 1.4624780416488647,
+      "learning_rate": 1.3e-06,
+      "loss": 2.06,
+      "step": 748
+    },
+    {
+      "epoch": 0.93625,
+      "grad_norm": 4.1101460456848145,
+      "learning_rate": 1.275e-06,
+      "loss": 3.6805,
+      "step": 749
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 1.3366179466247559,
+      "learning_rate": 1.25e-06,
+      "loss": 2.5252,
+      "step": 750
+    },
+    {
+      "epoch": 0.93875,
+      "grad_norm": 1.6742165088653564,
+      "learning_rate": 1.2250000000000001e-06,
+      "loss": 2.4546,
+      "step": 751
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.976860523223877,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.5712,
+      "step": 752
+    },
+    {
+      "epoch": 0.94125,
+      "grad_norm": 2.7342398166656494,
+      "learning_rate": 1.175e-06,
+      "loss": 2.8187,
+      "step": 753
+    },
+    {
+      "epoch": 0.9425,
+      "grad_norm": 1.6856083869934082,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 2.2563,
+      "step": 754
+    },
+    {
+      "epoch": 0.94375,
+      "grad_norm": 1.8753656148910522,
+      "learning_rate": 1.125e-06,
+      "loss": 2.6912,
+      "step": 755
+    },
+    {
+      "epoch": 0.945,
+      "grad_norm": 1.1665239334106445,
+      "learning_rate": 1.1e-06,
+      "loss": 2.57,
+      "step": 756
+    },
+    {
+      "epoch": 0.94625,
+      "grad_norm": 0.815209150314331,
+      "learning_rate": 1.075e-06,
+      "loss": 2.1256,
+      "step": 757
+    },
+    {
+      "epoch": 0.9475,
+      "grad_norm": 4.842214107513428,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 2.4971,
+      "step": 758
+    },
+    {
+      "epoch": 0.94875,
+      "grad_norm": 2.0265555381774902,
+      "learning_rate": 1.025e-06,
+      "loss": 2.134,
+      "step": 759
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.723125696182251,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.7601,
+      "step": 760
+    },
+    {
+      "epoch": 0.95125,
+      "grad_norm": 0.8349653482437134,
+      "learning_rate": 9.750000000000002e-07,
+      "loss": 2.0944,
+      "step": 761
+    },
+    {
+      "epoch": 0.9525,
+      "grad_norm": 2.7524056434631348,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 2.9711,
+      "step": 762
+    },
+    {
+      "epoch": 0.95375,
+      "grad_norm": 3.6746034622192383,
+      "learning_rate": 9.25e-07,
+      "loss": 2.6018,
+      "step": 763
+    },
+    {
+      "epoch": 0.955,
+      "grad_norm": 4.50912618637085,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.2356,
+      "step": 764
+    },
+    {
+      "epoch": 0.95625,
+      "grad_norm": 2.569899797439575,
+      "learning_rate": 8.75e-07,
+      "loss": 2.3839,
+      "step": 765
+    },
+    {
+      "epoch": 0.9575,
+      "grad_norm": 2.2566895484924316,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 3.053,
+      "step": 766
+    },
+    {
+      "epoch": 0.95875,
+      "grad_norm": 1.2487592697143555,
+      "learning_rate": 8.250000000000001e-07,
+      "loss": 2.8636,
+      "step": 767
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2288849353790283,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.9255,
+      "step": 768
+    },
+    {
+      "epoch": 0.96125,
+      "grad_norm": 2.1028974056243896,
+      "learning_rate": 7.750000000000001e-07,
+      "loss": 2.5616,
+      "step": 769
+    },
+    {
+      "epoch": 0.9625,
+      "grad_norm": 1.2116632461547852,
+      "learning_rate": 7.5e-07,
+      "loss": 2.56,
+      "step": 770
+    },
+    {
+      "epoch": 0.96375,
+      "grad_norm": 2.2025723457336426,
+      "learning_rate": 7.25e-07,
+      "loss": 3.1438,
+      "step": 771
+    },
+    {
+      "epoch": 0.965,
+      "grad_norm": 2.4728424549102783,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.478,
+      "step": 772
+    },
+    {
+      "epoch": 0.96625,
+      "grad_norm": 1.092957615852356,
+      "learning_rate": 6.750000000000001e-07,
+      "loss": 2.288,
+      "step": 773
+    },
+    {
+      "epoch": 0.9675,
+      "grad_norm": 1.0330168008804321,
+      "learning_rate": 6.5e-07,
+      "loss": 2.4745,
+      "step": 774
+    },
+    {
+      "epoch": 0.96875,
+      "grad_norm": 1.3029001951217651,
+      "learning_rate": 6.25e-07,
+      "loss": 2.3561,
+      "step": 775
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4613988399505615,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.4335,
+      "step": 776
+    },
+    {
+      "epoch": 0.97125,
+      "grad_norm": 1.6901812553405762,
+      "learning_rate": 5.750000000000001e-07,
+      "loss": 2.2362,
+      "step": 777
+    },
+    {
+      "epoch": 0.9725,
+      "grad_norm": 2.4732651710510254,
+      "learning_rate": 5.5e-07,
+      "loss": 2.6915,
+      "step": 778
+    },
+    {
+      "epoch": 0.97375,
+      "grad_norm": 1.1935354471206665,
+      "learning_rate": 5.250000000000001e-07,
+      "loss": 2.3641,
+      "step": 779
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 1.6522023677825928,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.3062,
+      "step": 780
+    },
+    {
+      "epoch": 0.97625,
+      "grad_norm": 1.62522554397583,
+      "learning_rate": 4.7500000000000006e-07,
+      "loss": 2.3801,
+      "step": 781
+    },
+    {
+      "epoch": 0.9775,
+      "grad_norm": 14.57507610321045,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 2.5558,
+      "step": 782
+    },
+    {
+      "epoch": 0.97875,
+      "grad_norm": 3.2062268257141113,
+      "learning_rate": 4.2500000000000006e-07,
+      "loss": 2.9463,
+      "step": 783
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.024563789367676,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.9367,
+      "step": 784
+    },
+    {
+      "epoch": 0.98125,
+      "grad_norm": 2.408194065093994,
+      "learning_rate": 3.75e-07,
+      "loss": 2.5923,
+      "step": 785
+    },
+    {
+      "epoch": 0.9825,
+      "grad_norm": 1.4996728897094727,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 2.7131,
+      "step": 786
+    },
+    {
+      "epoch": 0.98375,
+      "grad_norm": 0.4530770182609558,
+      "learning_rate": 3.25e-07,
+      "loss": 1.6141,
+      "step": 787
+    },
+    {
+      "epoch": 0.985,
+      "grad_norm": 1.9080215692520142,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.8874,
+      "step": 788
+    },
+    {
+      "epoch": 0.98625,
+      "grad_norm": 5.6946635246276855,
+      "learning_rate": 2.75e-07,
+      "loss": 3.0954,
+      "step": 789
+    },
+    {
+      "epoch": 0.9875,
+      "grad_norm": 1.8404529094696045,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 2.5155,
+      "step": 790
+    },
+    {
+      "epoch": 0.98875,
+      "grad_norm": 1.1621073484420776,
+      "learning_rate": 2.2500000000000002e-07,
+      "loss": 2.2214,
+      "step": 791
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.576001167297363,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.3722,
+      "step": 792
+    },
+    {
+      "epoch": 0.99125,
+      "grad_norm": 2.642176866531372,
+      "learning_rate": 1.7500000000000002e-07,
+      "loss": 3.2569,
+      "step": 793
+    },
+    {
+      "epoch": 0.9925,
+      "grad_norm": 2.1654837131500244,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 2.2922,
+      "step": 794
+    },
+    {
+      "epoch": 0.99375,
+      "grad_norm": 1.3945523500442505,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 2.2981,
+      "step": 795
+    },
+    {
+      "epoch": 0.995,
+      "grad_norm": 1.6085453033447266,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.4976,
+      "step": 796
+    },
+    {
+      "epoch": 0.99625,
+      "grad_norm": 3.1184237003326416,
+      "learning_rate": 7.500000000000001e-08,
+      "loss": 2.3142,
+      "step": 797
+    },
+    {
+      "epoch": 0.9975,
+      "grad_norm": 2.5621306896209717,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.8623,
+      "step": 798
+    },
+    {
+      "epoch": 0.99875,
+      "grad_norm": 3.707613468170166,
+      "learning_rate": 2.5000000000000002e-08,
+      "loss": 3.019,
+      "step": 799
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.426806926727295,
+      "learning_rate": 0.0,
+      "loss": 2.7174,
+      "step": 800
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 2.631765127182007,
+      "eval_runtime": 130.0102,
+      "eval_samples_per_second": 1.538,
+      "eval_steps_per_second": 1.538,
+      "step": 800
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 800,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.0586892345344e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}