diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7835 @@
+{
+  "best_global_step": 91000,
+  "best_metric": 3.857421875,
+  "best_model_checkpoint": "checkpoints/sonar_mini_5e4/checkpoint-91000",
+  "epoch": 1.0,
+  "eval_steps": 1000,
+  "global_step": 100000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001,
+      "grad_norm": 2.9780468940734863,
+      "learning_rate": 6.7437500000000004e-06,
+      "loss": 7.4228,
+      "step": 100
+    },
+    {
+      "epoch": 0.002,
+      "grad_norm": 0.274592787027359,
+      "learning_rate": 1.29875e-05,
+      "loss": 5.1812,
+      "step": 200
+    },
+    {
+      "epoch": 0.003,
+      "grad_norm": 0.12851963937282562,
+      "learning_rate": 1.923125e-05,
+      "loss": 4.6227,
+      "step": 300
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 0.0915624350309372,
+      "learning_rate": 2.5475000000000003e-05,
+      "loss": 4.4829,
+      "step": 400
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 0.08360039442777634,
+      "learning_rate": 3.1718749999999995e-05,
+      "loss": 4.4024,
+      "step": 500
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 0.08585469424724579,
+      "learning_rate": 3.7962499999999994e-05,
+      "loss": 4.3443,
+      "step": 600
+    },
+    {
+      "epoch": 0.007,
+      "grad_norm": 0.08225593715906143,
+      "learning_rate": 4.4206249999999994e-05,
+      "loss": 4.3023,
+      "step": 700
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 0.09392552822828293,
+      "learning_rate": 5.045e-05,
+      "loss": 4.2707,
+      "step": 800
+    },
+    {
+      "epoch": 0.009,
+      "grad_norm": 0.08779512345790863,
+      "learning_rate": 5.669375e-05,
+      "loss": 4.2444,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.09750315546989441,
+      "learning_rate": 6.293749999999999e-05,
+      "loss": 4.22,
+      "step": 1000
+    },
+    {
+      "epoch": 0.011,
+      "grad_norm": 0.08571433275938034,
+      "learning_rate": 6.918125e-05,
+      "loss": 4.2002,
+      "step": 1100
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 0.09333999454975128,
+      "learning_rate": 7.542499999999999e-05,
+      "loss": 4.1821,
+      "step": 1200
+    },
+    {
+      "epoch": 0.013,
+      "grad_norm": 0.09406431764364243,
+      "learning_rate": 8.166875e-05,
+      "loss": 4.166,
+      "step": 1300
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 0.08803971856832504,
+      "learning_rate": 8.791249999999999e-05,
+      "loss": 4.1483,
+      "step": 1400
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 0.10886548459529877,
+      "learning_rate": 9.415625e-05,
+      "loss": 4.1324,
+      "step": 1500
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.10178254544734955,
+      "learning_rate": 0.0001004,
+      "loss": 4.1196,
+      "step": 1600
+    },
+    {
+      "epoch": 0.017,
+      "grad_norm": 0.09580868482589722,
+      "learning_rate": 0.00010664375,
+      "loss": 4.107,
+      "step": 1700
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 0.0989207997918129,
+      "learning_rate": 0.0001128875,
+      "loss": 4.0939,
+      "step": 1800
+    },
+    {
+      "epoch": 0.019,
+      "grad_norm": 0.08986181765794754,
+      "learning_rate": 0.00011913125,
+      "loss": 4.0821,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.10257107019424438,
+      "learning_rate": 0.000125375,
+      "loss": 4.0721,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 4.4296875,
+      "eval_runtime": 33.6861,
+      "eval_samples_per_second": 17876.479,
+      "eval_steps_per_second": 17.485,
+      "step": 2000
+    },
+    {
+      "epoch": 0.021,
+      "grad_norm": 0.10034991055727005,
+      "learning_rate": 0.00013161875,
+      "loss": 4.0629,
+      "step": 2100
+    },
+    {
+      "epoch": 0.022,
+      "grad_norm": 0.10683776438236237,
+      "learning_rate": 0.0001378625,
+      "loss": 4.0515,
+      "step": 2200
+    },
+    {
+      "epoch": 0.023,
+      "grad_norm": 0.10092870891094208,
+      "learning_rate": 0.00014410625,
+      "loss": 4.0428,
+      "step": 2300
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 0.11656210571527481,
+      "learning_rate": 0.00015035,
+      "loss": 4.0305,
+      "step": 2400
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 0.10047990083694458,
+      "learning_rate": 0.00015659375,
+      "loss": 4.0236,
+      "step": 2500
+    },
+    {
+      "epoch": 0.026,
+      "grad_norm": 0.10436904430389404,
+      "learning_rate": 0.0001628375,
+      "loss": 4.014,
+      "step": 2600
+    },
+    {
+      "epoch": 0.027,
+      "grad_norm": 0.09785196930170059,
+      "learning_rate": 0.00016908125000000002,
+      "loss": 4.0062,
+      "step": 2700
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 0.10417915135622025,
+      "learning_rate": 0.000175325,
+      "loss": 3.9974,
+      "step": 2800
+    },
+    {
+      "epoch": 0.029,
+      "grad_norm": 0.10269144922494888,
+      "learning_rate": 0.00018156875,
+      "loss": 3.99,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.10645649582147598,
+      "learning_rate": 0.0001878125,
+      "loss": 3.9836,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.3515625,
+      "eval_runtime": 33.1738,
+      "eval_samples_per_second": 18152.539,
+      "eval_steps_per_second": 17.755,
+      "step": 3000
+    },
+    {
+      "epoch": 0.031,
+      "grad_norm": 0.11746224015951157,
+      "learning_rate": 0.00019405625000000002,
+      "loss": 3.9753,
+      "step": 3100
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.10270912200212479,
+      "learning_rate": 0.00020030000000000002,
+      "loss": 3.9693,
+      "step": 3200
+    },
+    {
+      "epoch": 0.033,
+      "grad_norm": 0.10418165475130081,
+      "learning_rate": 0.00020654375,
+      "loss": 3.9633,
+      "step": 3300
+    },
+    {
+      "epoch": 0.034,
+      "grad_norm": 0.10649538040161133,
+      "learning_rate": 0.0002127875,
+      "loss": 3.9519,
+      "step": 3400
+    },
+    {
+      "epoch": 0.035,
+      "grad_norm": 0.11068189144134521,
+      "learning_rate": 0.00021903125000000001,
+      "loss": 3.9479,
+      "step": 3500
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 0.11896345019340515,
+      "learning_rate": 0.00022527500000000001,
+      "loss": 3.9408,
+      "step": 3600
+    },
+    {
+      "epoch": 0.037,
+      "grad_norm": 0.09746250510215759,
+      "learning_rate": 0.00023151875000000004,
+      "loss": 3.9353,
+      "step": 3700
+    },
+    {
+      "epoch": 0.038,
+      "grad_norm": 0.15779097378253937,
+      "learning_rate": 0.00023776249999999999,
+      "loss": 3.9304,
+      "step": 3800
+    },
+    {
+      "epoch": 0.039,
+      "grad_norm": 0.10786330699920654,
+      "learning_rate": 0.00024400625,
+      "loss": 3.9227,
+      "step": 3900
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.11500537395477295,
+      "learning_rate": 0.00025025,
+      "loss": 3.9166,
+      "step": 4000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 4.296875,
+      "eval_runtime": 33.1436,
+      "eval_samples_per_second": 18169.071,
+      "eval_steps_per_second": 17.771,
+      "step": 4000
+    },
+    {
+      "epoch": 0.041,
+      "grad_norm": 0.13655492663383484,
+      "learning_rate": 0.00025649375,
+      "loss": 3.9107,
+      "step": 4100
+    },
+    {
+      "epoch": 0.042,
+      "grad_norm": 0.10026777535676956,
+      "learning_rate": 0.00026273750000000004,
+      "loss": 3.9054,
+      "step": 4200
+    },
+    {
+      "epoch": 0.043,
+      "grad_norm": 0.12338188290596008,
+      "learning_rate": 0.00026898125000000004,
+      "loss": 3.9006,
+      "step": 4300
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 0.10768528282642365,
+      "learning_rate": 0.000275225,
+      "loss": 3.894,
+      "step": 4400
+    },
+    {
+      "epoch": 0.045,
+      "grad_norm": 0.10258147865533829,
+      "learning_rate": 0.00028146875,
+      "loss": 3.8912,
+      "step": 4500
+    },
+    {
+      "epoch": 0.046,
+      "grad_norm": 0.17049455642700195,
+      "learning_rate": 0.0002877125,
+      "loss": 3.8824,
+      "step": 4600
+    },
+    {
+      "epoch": 0.047,
+      "grad_norm": 0.10924670845270157,
+      "learning_rate": 0.00029395625000000003,
+      "loss": 3.8803,
+      "step": 4700
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.12295184284448624,
+      "learning_rate": 0.0003002,
+      "loss": 3.8727,
+      "step": 4800
+    },
+    {
+      "epoch": 0.049,
+      "grad_norm": 0.1435551792383194,
+      "learning_rate": 0.00030644375000000003,
+      "loss": 3.8705,
+      "step": 4900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.16839157044887543,
+      "learning_rate": 0.00031268750000000003,
+      "loss": 3.8646,
+      "step": 5000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 4.25,
+      "eval_runtime": 33.1236,
+      "eval_samples_per_second": 18180.027,
+      "eval_steps_per_second": 17.782,
+      "step": 5000
+    },
+    {
+      "epoch": 0.051,
+      "grad_norm": 0.12637090682983398,
+      "learning_rate": 0.00031893125000000003,
+      "loss": 3.86,
+      "step": 5100
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 0.16098865866661072,
+      "learning_rate": 0.000325175,
+      "loss": 3.8559,
+      "step": 5200
+    },
+    {
+      "epoch": 0.053,
+      "grad_norm": 0.10894348472356796,
+      "learning_rate": 0.00033141875,
+      "loss": 3.8512,
+      "step": 5300
+    },
+    {
+      "epoch": 0.054,
+      "grad_norm": 0.11462244391441345,
+      "learning_rate": 0.00033766250000000003,
+      "loss": 3.8469,
+      "step": 5400
+    },
+    {
+      "epoch": 0.055,
+      "grad_norm": 0.111247718334198,
+      "learning_rate": 0.00034390625000000003,
+      "loss": 3.842,
+      "step": 5500
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 0.14288830757141113,
+      "learning_rate": 0.00035015,
+      "loss": 3.8373,
+      "step": 5600
+    },
+    {
+      "epoch": 0.057,
+      "grad_norm": 0.14327366650104523,
+      "learning_rate": 0.00035639375,
+      "loss": 3.8346,
+      "step": 5700
+    },
+    {
+      "epoch": 0.058,
+      "grad_norm": 0.10640984773635864,
+      "learning_rate": 0.0003626375,
+      "loss": 3.8318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.059,
+      "grad_norm": 0.12228193134069443,
+      "learning_rate": 0.0003688812500000001,
+      "loss": 3.8274,
+      "step": 5900
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.12095323950052261,
+      "learning_rate": 0.00037512499999999997,
+      "loss": 3.8233,
+      "step": 6000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 4.21875,
+      "eval_runtime": 33.1908,
+      "eval_samples_per_second": 18143.214,
+      "eval_steps_per_second": 17.746,
+      "step": 6000
+    },
+    {
+      "epoch": 0.061,
+      "grad_norm": 0.17427460849285126,
+      "learning_rate": 0.00038136874999999997,
+      "loss": 3.8206,
+      "step": 6100
+    },
+    {
+      "epoch": 0.062,
+      "grad_norm": 0.17270584404468536,
+      "learning_rate": 0.0003876125,
+      "loss": 3.8141,
+      "step": 6200
+    },
+    {
+      "epoch": 0.063,
+      "grad_norm": 0.10496606677770615,
+      "learning_rate": 0.00039385624999999997,
+      "loss": 3.8104,
+      "step": 6300
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.18062132596969604,
+      "learning_rate": 0.0004001,
+      "loss": 3.8084,
+      "step": 6400
+    },
+    {
+      "epoch": 0.065,
+      "grad_norm": 0.1837543398141861,
+      "learning_rate": 0.00040634375,
+      "loss": 3.8042,
+      "step": 6500
+    },
+    {
+      "epoch": 0.066,
+      "grad_norm": 0.1967838704586029,
+      "learning_rate": 0.0004125875,
+      "loss": 3.7992,
+      "step": 6600
+    },
+    {
+      "epoch": 0.067,
+      "grad_norm": 0.23508614301681519,
+      "learning_rate": 0.00041883125,
+      "loss": 3.7988,
+      "step": 6700
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 0.15716159343719482,
+      "learning_rate": 0.00042507499999999996,
+      "loss": 3.7925,
+      "step": 6800
+    },
+    {
+      "epoch": 0.069,
+      "grad_norm": 0.22673483192920685,
+      "learning_rate": 0.00043131875,
+      "loss": 3.7906,
+      "step": 6900
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2622189521789551,
+      "learning_rate": 0.0004375625,
+      "loss": 3.7868,
+      "step": 7000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 4.19921875,
+      "eval_runtime": 33.1108,
+      "eval_samples_per_second": 18187.079,
+      "eval_steps_per_second": 17.789,
+      "step": 7000
+    },
+    {
+      "epoch": 0.071,
+      "grad_norm": 0.12971626222133636,
+      "learning_rate": 0.00044380624999999996,
+      "loss": 3.7834,
+      "step": 7100
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 0.12215161323547363,
+      "learning_rate": 0.00045005,
+      "loss": 3.7807,
+      "step": 7200
+    },
+    {
+      "epoch": 0.073,
+      "grad_norm": 0.17193715274333954,
+      "learning_rate": 0.00045629375,
+      "loss": 3.7767,
+      "step": 7300
+    },
+    {
+      "epoch": 0.074,
+      "grad_norm": 0.14749088883399963,
+      "learning_rate": 0.00046253750000000007,
+      "loss": 3.7741,
+      "step": 7400
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 0.21286271512508392,
+      "learning_rate": 0.00046878125,
+      "loss": 3.771,
+      "step": 7500
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 0.2201075553894043,
+      "learning_rate": 0.00047502499999999996,
+      "loss": 3.7676,
+      "step": 7600
+    },
+    {
+      "epoch": 0.077,
+      "grad_norm": 0.20476743578910828,
+      "learning_rate": 0.00048126875,
+      "loss": 3.7648,
+      "step": 7700
+    },
+    {
+      "epoch": 0.078,
+      "grad_norm": 0.1267070174217224,
+      "learning_rate": 0.0004875125,
+      "loss": 3.7634,
+      "step": 7800
+    },
+    {
+      "epoch": 0.079,
+      "grad_norm": 0.13982422649860382,
+      "learning_rate": 0.00049375625,
+      "loss": 3.7585,
+      "step": 7900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.17995713651180267,
+      "learning_rate": 0.0005,
+      "loss": 3.7567,
+      "step": 8000
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 4.15625,
+      "eval_runtime": 33.1093,
+      "eval_samples_per_second": 18187.899,
+      "eval_steps_per_second": 17.79,
+      "step": 8000
+    },
+    {
+      "epoch": 0.081,
+      "grad_norm": 0.14613783359527588,
+      "learning_rate": 0.0004969039949999533,
+      "loss": 3.7543,
+      "step": 8100
+    },
+    {
+      "epoch": 0.082,
+      "grad_norm": 0.13322409987449646,
+      "learning_rate": 0.0004938647983247948,
+      "loss": 3.7495,
+      "step": 8200
+    },
+    {
+      "epoch": 0.083,
+      "grad_norm": 0.23195946216583252,
+      "learning_rate": 0.0004908806936738159,
+      "loss": 3.7483,
+      "step": 8300
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 0.22853663563728333,
+      "learning_rate": 0.00048795003647426655,
+      "loss": 3.7421,
+      "step": 8400
+    },
+    {
+      "epoch": 0.085,
+      "grad_norm": 0.19258145987987518,
+      "learning_rate": 0.0004850712500726659,
+      "loss": 3.74,
+      "step": 8500
+    },
+    {
+      "epoch": 0.086,
+      "grad_norm": 0.15334059298038483,
+      "learning_rate": 0.0004822428221704122,
+      "loss": 3.7385,
+      "step": 8600
+    },
+    {
+      "epoch": 0.087,
+      "grad_norm": 0.13609085977077484,
+      "learning_rate": 0.0004794633014853842,
+      "loss": 3.7364,
+      "step": 8700
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 0.16103780269622803,
+      "learning_rate": 0.00047673129462279614,
+      "loss": 3.7331,
+      "step": 8800
+    },
+    {
+      "epoch": 0.089,
+      "grad_norm": 0.13547872006893158,
+      "learning_rate": 0.00047404546313997725,
+      "loss": 3.7288,
+      "step": 8900
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1519702523946762,
+      "learning_rate": 0.00047140452079103175,
+      "loss": 3.7258,
+      "step": 9000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 4.13671875,
+      "eval_runtime": 33.1497,
+      "eval_samples_per_second": 18165.735,
+      "eval_steps_per_second": 17.768,
+      "step": 9000
+    },
+    {
+      "epoch": 0.091,
+      "grad_norm": 0.13400638103485107,
+      "learning_rate": 0.00046880723093849544,
+      "loss": 3.7264,
+      "step": 9100
+    },
+    {
+      "epoch": 0.092,
+      "grad_norm": 0.17761798202991486,
+      "learning_rate": 0.0004662524041201569,
+      "loss": 3.7228,
+      "step": 9200
+    },
+    {
+      "epoch": 0.093,
+      "grad_norm": 0.10788439959287643,
+      "learning_rate": 0.00046373889576016824,
+      "loss": 3.7187,
+      "step": 9300
+    },
+    {
+      "epoch": 0.094,
+      "grad_norm": 0.13771696388721466,
+      "learning_rate": 0.00046126560401444256,
+      "loss": 3.7147,
+      "step": 9400
+    },
+    {
+      "epoch": 0.095,
+      "grad_norm": 0.1324366182088852,
+      "learning_rate": 0.0004588314677411235,
+      "loss": 3.7148,
+      "step": 9500
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.15091444551944733,
+      "learning_rate": 0.0004564354645876385,
+      "loss": 3.7135,
+      "step": 9600
+    },
+    {
+      "epoch": 0.097,
+      "grad_norm": 0.1656925231218338,
+      "learning_rate": 0.0004540766091864998,
+      "loss": 3.7098,
+      "step": 9700
+    },
+    {
+      "epoch": 0.098,
+      "grad_norm": 0.13978664577007294,
+      "learning_rate": 0.00045175395145262567,
+      "loss": 3.7062,
+      "step": 9800
+    },
+    {
+      "epoch": 0.099,
+      "grad_norm": 0.10810708999633789,
+      "learning_rate": 0.0004494665749754947,
+      "loss": 3.7057,
+      "step": 9900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1327275186777115,
+      "learning_rate": 0.00044721359549995795,
+      "loss": 3.7022,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 4.11328125,
+      "eval_runtime": 33.1707,
+      "eval_samples_per_second": 18154.235,
+      "eval_steps_per_second": 17.757,
+      "step": 10000
+    },
+    {
+      "epoch": 0.101,
+      "grad_norm": 0.1312229484319687,
+      "learning_rate": 0.0004449941594899848,
+      "loss": 3.7018,
+      "step": 10100
+    },
+    {
+      "epoch": 0.102,
+      "grad_norm": 0.12260957062244415,
+      "learning_rate": 0.0004428074427700477,
+      "loss": 3.6961,
+      "step": 10200
+    },
+    {
+      "epoch": 0.103,
+      "grad_norm": 0.10720884054899216,
+      "learning_rate": 0.0004406526492392317,
+      "loss": 3.6958,
+      "step": 10300
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 0.10942485928535461,
+      "learning_rate": 0.0004385290096535146,
+      "loss": 3.6949,
+      "step": 10400
+    },
+    {
+      "epoch": 0.105,
+      "grad_norm": 0.11342720687389374,
+      "learning_rate": 0.0004364357804719848,
+      "loss": 3.6954,
+      "step": 10500
+    },
+    {
+      "epoch": 0.106,
+      "grad_norm": 0.12471634149551392,
+      "learning_rate": 0.0004343722427630694,
+      "loss": 3.6926,
+      "step": 10600
+    },
+    {
+      "epoch": 0.107,
+      "grad_norm": 0.11318652331829071,
+      "learning_rate": 0.00043233770116711697,
+      "loss": 3.6912,
+      "step": 10700
+    },
+    {
+      "epoch": 0.108,
+      "grad_norm": 0.11893333494663239,
+      "learning_rate": 0.0004303314829119352,
+      "loss": 3.6893,
+      "step": 10800
+    },
+    {
+      "epoch": 0.109,
+      "grad_norm": 0.12605136632919312,
+      "learning_rate": 0.0004283529368781193,
+      "loss": 3.6862,
+      "step": 10900
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.24888132512569427,
+      "learning_rate": 0.00042640143271122083,
+      "loss": 3.6851,
+      "step": 11000
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 4.09765625,
+      "eval_runtime": 33.0869,
+      "eval_samples_per_second": 18200.176,
+      "eval_steps_per_second": 17.802,
+      "step": 11000
+    },
+    {
+      "epoch": 0.111,
+      "grad_norm": 0.1126684695482254,
+      "learning_rate": 0.00042447635997800896,
+      "loss": 3.6798,
+      "step": 11100
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.16141295433044434,
+      "learning_rate": 0.0004225771273642583,
+      "loss": 3.6806,
+      "step": 11200
+    },
+    {
+      "epoch": 0.113,
+      "grad_norm": 0.18238785862922668,
+      "learning_rate": 0.0004207031619116712,
+      "loss": 3.6773,
+      "step": 11300
+    },
+    {
+      "epoch": 0.114,
+      "grad_norm": 0.14396843314170837,
+      "learning_rate": 0.0004188539082916955,
+      "loss": 3.6765,
+      "step": 11400
+    },
+    {
+      "epoch": 0.115,
+      "grad_norm": 0.20065593719482422,
+      "learning_rate": 0.0004170288281141495,
+      "loss": 3.673,
+      "step": 11500
+    },
+    {
+      "epoch": 0.116,
+      "grad_norm": 0.12579910457134247,
+      "learning_rate": 0.00041522739926869986,
+      "loss": 3.6727,
+      "step": 11600
+    },
+    {
+      "epoch": 0.117,
+      "grad_norm": 0.10580720007419586,
+      "learning_rate": 0.0004134491152973616,
+      "loss": 3.6699,
+      "step": 11700
+    },
+    {
+      "epoch": 0.118,
+      "grad_norm": 0.12733343243598938,
+      "learning_rate": 0.00041169348479630914,
+      "loss": 3.6685,
+      "step": 11800
+    },
+    {
+      "epoch": 0.119,
+      "grad_norm": 0.09983912110328674,
+      "learning_rate": 0.0004099600308453939,
+      "loss": 3.6677,
+      "step": 11900
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1023598462343216,
+      "learning_rate": 0.0004082482904638631,
+      "loss": 3.6669,
+      "step": 12000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 4.08203125,
+      "eval_runtime": 33.0923,
+      "eval_samples_per_second": 18197.203,
+      "eval_steps_per_second": 17.799,
+      "step": 12000
+    },
+    {
+      "epoch": 0.121,
+      "grad_norm": 0.12581580877304077,
+      "learning_rate": 0.0004065578140908709,
+      "loss": 3.6641,
+      "step": 12100
+    },
+    {
+      "epoch": 0.122,
+      "grad_norm": 0.11037498712539673,
+      "learning_rate": 0.000404888165089458,
+      "loss": 3.664,
+      "step": 12200
+    },
+    {
+      "epoch": 0.123,
+      "grad_norm": 0.11614446341991425,
+      "learning_rate": 0.0004032389192727559,
+      "loss": 3.6598,
+      "step": 12300
+    },
+    {
+      "epoch": 0.124,
+      "grad_norm": 0.12279075384140015,
+      "learning_rate": 0.0004016096644512495,
+      "loss": 3.6603,
+      "step": 12400
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.1049276739358902,
+      "learning_rate": 0.0004,
+      "loss": 3.6569,
+      "step": 12500
+    },
+    {
+      "epoch": 0.126,
+      "grad_norm": 0.16340377926826477,
+      "learning_rate": 0.0003984095364447979,
+      "loss": 3.6581,
+      "step": 12600
+    },
+    {
+      "epoch": 0.127,
+      "grad_norm": 0.10878612846136093,
+      "learning_rate": 0.00039683789506627255,
+      "loss": 3.6548,
+      "step": 12700
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.11425680667161942,
+      "learning_rate": 0.00039528470752104737,
+      "loss": 3.6564,
+      "step": 12800
+    },
+    {
+      "epoch": 0.129,
+      "grad_norm": 0.12371216714382172,
+      "learning_rate": 0.0003937496154790789,
+      "loss": 3.6509,
+      "step": 12900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1376781165599823,
+      "learning_rate": 0.0003922322702763681,
+      "loss": 3.6531,
+      "step": 13000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 4.0625,
+      "eval_runtime": 33.0799,
+      "eval_samples_per_second": 18204.035,
+      "eval_steps_per_second": 17.805,
+      "step": 13000
+    },
+    {
+      "epoch": 0.131,
+      "grad_norm": 0.15412083268165588,
+      "learning_rate": 0.00039073233258228173,
+      "loss": 3.6495,
+      "step": 13100
+    },
+    {
+      "epoch": 0.132,
+      "grad_norm": 0.10344259440898895,
+      "learning_rate": 0.0003892494720807615,
+      "loss": 3.652,
+      "step": 13200
+    },
+    {
+      "epoch": 0.133,
+      "grad_norm": 0.12037135660648346,
+      "learning_rate": 0.0003877833671647406,
+      "loss": 3.6492,
+      "step": 13300
+    },
+    {
+      "epoch": 0.134,
+      "grad_norm": 0.10815135389566422,
+      "learning_rate": 0.0003863337046431279,
+      "loss": 3.6461,
+      "step": 13400
+    },
+    {
+      "epoch": 0.135,
+      "grad_norm": 0.14830388128757477,
+      "learning_rate": 0.00038490017945975053,
+      "loss": 3.6432,
+      "step": 13500
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 0.14780588448047638,
+      "learning_rate": 0.0003834824944236852,
+      "loss": 3.641,
+      "step": 13600
+    },
+    {
+      "epoch": 0.137,
+      "grad_norm": 0.13959766924381256,
+      "learning_rate": 0.00038208035995043505,
+      "loss": 3.6423,
+      "step": 13700
+    },
+    {
+      "epoch": 0.138,
+      "grad_norm": 0.1739835888147354,
+      "learning_rate": 0.0003806934938134405,
+      "loss": 3.6412,
+      "step": 13800
+    },
+    {
+      "epoch": 0.139,
+      "grad_norm": 0.1504901498556137,
+      "learning_rate": 0.0003793216209054408,
+      "loss": 3.6402,
+      "step": 13900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.11134779453277588,
+      "learning_rate": 0.0003779644730092272,
+      "loss": 3.6379,
+      "step": 14000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 4.0546875,
+      "eval_runtime": 33.0435,
+      "eval_samples_per_second": 18224.12,
+      "eval_steps_per_second": 17.825,
+      "step": 14000
+    },
+    {
+      "epoch": 0.141,
+      "grad_norm": 0.14684522151947021,
+      "learning_rate": 0.0003766217885773547,
+      "loss": 3.638,
+      "step": 14100
+    },
+    {
+      "epoch": 0.142,
+      "grad_norm": 0.10334658622741699,
+      "learning_rate": 0.0003752933125204008,
+      "loss": 3.6364,
+      "step": 14200
+    },
+    {
+      "epoch": 0.143,
+      "grad_norm": 0.1433073729276657,
+      "learning_rate": 0.00037397879600338285,
+      "loss": 3.636,
+      "step": 14300
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.112115278840065,
+      "learning_rate": 0.000372677996249965,
+      "loss": 3.6345,
+      "step": 14400
+    },
+    {
+      "epoch": 0.145,
+      "grad_norm": 0.11925119906663895,
+      "learning_rate": 0.0003713906763541037,
+      "loss": 3.6319,
+      "step": 14500
+    },
+    {
+      "epoch": 0.146,
+      "grad_norm": 0.12104904651641846,
+      "learning_rate": 0.00037011660509880264,
+      "loss": 3.6304,
+      "step": 14600
+    },
+    {
+      "epoch": 0.147,
+      "grad_norm": 0.13862280547618866,
+      "learning_rate": 0.0003688555567816587,
+      "loss": 3.6297,
+      "step": 14700
+    },
+    {
+      "epoch": 0.148,
+      "grad_norm": 0.1608782410621643,
+      "learning_rate": 0.00036760731104690394,
+      "loss": 3.6296,
+      "step": 14800
+    },
+    {
+      "epoch": 0.149,
+      "grad_norm": 0.09898547828197479,
+      "learning_rate": 0.0003663716527236559,
+      "loss": 3.6265,
+      "step": 14900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.11533054709434509,
+      "learning_rate": 0.00036514837167011074,
+      "loss": 3.6266,
+      "step": 15000
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 4.05078125,
+      "eval_runtime": 33.0573,
+      "eval_samples_per_second": 18216.467,
+      "eval_steps_per_second": 17.818,
+      "step": 15000
+    },
+    {
+      "epoch": 0.151,
+      "grad_norm": 0.10128284990787506,
+      "learning_rate": 0.0003639372626234195,
+      "loss": 3.6259,
+      "step": 15100
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 0.12907330691814423,
+      "learning_rate": 0.0003627381250550059,
+      "loss": 3.6277,
+      "step": 15200
+    },
+    {
+      "epoch": 0.153,
+      "grad_norm": 0.09283538907766342,
+      "learning_rate": 0.0003615507630310936,
+      "loss": 3.6219,
+      "step": 15300
+    },
+    {
+      "epoch": 0.154,
+      "grad_norm": 0.10961660742759705,
+      "learning_rate": 0.0003603749850782236,
+      "loss": 3.6233,
+      "step": 15400
+    },
+    {
+      "epoch": 0.155,
+      "grad_norm": 0.11972799897193909,
+      "learning_rate": 0.00035921060405354985,
+      "loss": 3.6202,
+      "step": 15500
+    },
+    {
+      "epoch": 0.156,
+      "grad_norm": 0.11016960442066193,
+      "learning_rate": 0.0003580574370197164,
+      "loss": 3.6204,
+      "step": 15600
+    },
+    {
+      "epoch": 0.157,
+      "grad_norm": 0.11285857856273651,
+      "learning_rate": 0.0003569153051241248,
+      "loss": 3.6205,
+      "step": 15700
+    },
+    {
+      "epoch": 0.158,
+      "grad_norm": 0.11873608082532883,
+      "learning_rate": 0.00035578403348241,
+      "loss": 3.619,
+      "step": 15800
+    },
+    {
+      "epoch": 0.159,
+      "grad_norm": 0.1388588696718216,
+      "learning_rate": 0.0003546634510659543,
+      "loss": 3.6178,
+      "step": 15900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.10858651995658875,
+      "learning_rate": 0.00035355339059327376,
+      "loss": 3.6183,
+      "step": 16000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 4.03515625,
+      "eval_runtime": 33.0196,
+      "eval_samples_per_second": 18237.308,
+      "eval_steps_per_second": 17.838,
+      "step": 16000
+    },
+    {
+      "epoch": 0.161,
+      "grad_norm": 0.09829683601856232,
+      "learning_rate": 0.0003524536884251207,
+      "loss": 3.618,
+      "step": 16100
+    },
+    {
+      "epoch": 0.162,
+      "grad_norm": 0.12840525805950165,
+      "learning_rate": 0.00035136418446315325,
+      "loss": 3.6153,
+      "step": 16200
+    },
+    {
+      "epoch": 0.163,
+      "grad_norm": 0.13450373709201813,
+      "learning_rate": 0.00035028472205202936,
+      "loss": 3.6142,
+      "step": 16300
+    },
+    {
+      "epoch": 0.164,
+      "grad_norm": 0.09422667324542999,
+      "learning_rate": 0.00034921514788478915,
+      "loss": 3.6129,
+      "step": 16400
+    },
+    {
+      "epoch": 0.165,
+      "grad_norm": 0.1105048656463623,
+      "learning_rate": 0.0003481553119113957,
+      "loss": 3.6126,
+      "step": 16500
+    },
+    {
+      "epoch": 0.166,
+      "grad_norm": 0.11182115972042084,
+      "learning_rate": 0.00034710506725031166,
+      "loss": 3.6077,
+      "step": 16600
+    },
+    {
+      "epoch": 0.167,
+      "grad_norm": 0.15074919164180756,
+      "learning_rate": 0.0003460642701029914,
+      "loss": 3.6093,
+      "step": 16700
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 0.11445897072553635,
+      "learning_rate": 0.00034503277967117707,
+      "loss": 3.6077,
+      "step": 16800
+    },
+    {
+      "epoch": 0.169,
+      "grad_norm": 0.10784851759672165,
+      "learning_rate": 0.0003440104580768907,
+      "loss": 3.6088,
+      "step": 16900
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.11717010289430618,
+      "learning_rate": 0.00034299717028501764,
+      "loss": 3.6059,
+      "step": 17000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 4.03125,
+      "eval_runtime": 33.0161,
+      "eval_samples_per_second": 18239.208,
+      "eval_steps_per_second": 17.84,
+      "step": 17000
+    },
+    {
+      "epoch": 0.171,
+      "grad_norm": 0.10456647723913193,
+      "learning_rate": 0.00034199278402838475,
+      "loss": 3.6086,
+      "step": 17100
+    },
+    {
+      "epoch": 0.172,
+      "grad_norm": 0.11512298136949539,
+      "learning_rate": 0.00034099716973523677,
+      "loss": 3.6081,
+      "step": 17200
+    },
+    {
+      "epoch": 0.173,
+      "grad_norm": 0.11558009684085846,
+      "learning_rate": 0.000340010200459023,
+      "loss": 3.605,
+      "step": 17300
+    },
+    {
+      "epoch": 0.174,
+      "grad_norm": 0.12961934506893158,
+      "learning_rate": 0.0003390317518104052,
+      "loss": 3.6042,
+      "step": 17400
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 0.10916338860988617,
+      "learning_rate": 0.0003380617018914066,
+      "loss": 3.605,
+      "step": 17500
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.09793013334274292,
+      "learning_rate": 0.00033709993123162105,
+      "loss": 3.6012,
+      "step": 17600
+    },
+    {
+      "epoch": 0.177,
+      "grad_norm": 0.09620773047208786,
+      "learning_rate": 0.0003361463227264072,
+      "loss": 3.6022,
+      "step": 17700
+    },
+    {
+      "epoch": 0.178,
+      "grad_norm": 0.1438409984111786,
+      "learning_rate": 0.0003352007615769955,
+      "loss": 3.5989,
+      "step": 17800
+    },
+    {
+      "epoch": 0.179,
+      "grad_norm": 0.09917107224464417,
+      "learning_rate": 0.0003342631352324378,
+      "loss": 3.5997,
+      "step": 17900
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1120196059346199,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 3.5982,
+      "step": 18000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 4.02734375,
+      "eval_runtime": 32.9951,
+      "eval_samples_per_second": 18250.853,
+      "eval_steps_per_second": 17.851,
+      "step": 18000
+    },
+    {
+      "epoch": 0.181,
+      "grad_norm": 0.11737548559904099,
+      "learning_rate": 0.00033241124765726683,
+      "loss": 3.6,
+      "step": 18100
+    },
+    {
+      "epoch": 0.182,
+      "grad_norm": 0.09610620886087418,
+      "learning_rate": 0.00033149677206589795,
+      "loss": 3.5976,
+      "step": 18200
+    },
+    {
+      "epoch": 0.183,
+      "grad_norm": 0.0966159924864769,
+      "learning_rate": 0.00033058980245364314,
+      "loss": 3.5956,
+      "step": 18300
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 0.12272274494171143,
+      "learning_rate": 0.00032969023669789354,
+      "loss": 3.5948,
+      "step": 18400
+    },
+    {
+      "epoch": 0.185,
+      "grad_norm": 0.12435866892337799,
+      "learning_rate": 0.0003287979746107146,
+      "loss": 3.5962,
+      "step": 18500
+    },
+    {
+      "epoch": 0.186,
+      "grad_norm": 0.13750095665454865,
+      "learning_rate": 0.0003279129178919765,
+      "loss": 3.5942,
+      "step": 18600
+    },
+    {
+      "epoch": 0.187,
+      "grad_norm": 0.09951788187026978,
+      "learning_rate": 0.00032703497008386434,
+      "loss": 3.5948,
+      "step": 18700
+    },
+    {
+      "epoch": 0.188,
+      "grad_norm": 0.17174014449119568,
+      "learning_rate": 0.0003261640365267211,
+      "loss": 3.5918,
+      "step": 18800
+    },
+    {
+      "epoch": 0.189,
+      "grad_norm": 0.09349057078361511,
+      "learning_rate": 0.0003253000243161777,
+      "loss": 3.5933,
+      "step": 18900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13002406060695648,
+      "learning_rate": 0.0003244428422615251,
+      "loss": 3.5934,
+      "step": 19000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 4.01171875,
+      "eval_runtime": 33.1166,
+      "eval_samples_per_second": 18183.864,
+      "eval_steps_per_second": 17.786,
+      "step": 19000
+    },
+    {
+      "epoch": 0.191,
+      "grad_norm": 0.09189624339342117,
+      "learning_rate": 0.0003235924008452868,
+      "loss": 3.5931,
+      "step": 19100
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.14783063530921936,
+      "learning_rate": 0.0003227486121839514,
+      "loss": 3.5908,
+      "step": 19200
+    },
+    {
+      "epoch": 0.193,
+      "grad_norm": 0.12302859127521515,
+      "learning_rate": 0.00032191138998982524,
+      "loss": 3.59,
+      "step": 19300
+    },
+    {
+      "epoch": 0.194,
+      "grad_norm": 0.11037351936101913,
+      "learning_rate": 0.0003210806495339678,
+      "loss": 3.5881,
+      "step": 19400
+    },
+    {
+      "epoch": 0.195,
+      "grad_norm": 0.10322829335927963,
+      "learning_rate": 0.00032025630761017425,
+      "loss": 3.5877,
+      "step": 19500
+    },
+    {
+      "epoch": 0.196,
+      "grad_norm": 0.09346406161785126,
+      "learning_rate": 0.00031943828249997,
+      "loss": 3.5847,
+      "step": 19600
+    },
+    {
+      "epoch": 0.197,
+      "grad_norm": 0.10106361657381058,
+      "learning_rate": 0.0003186264939385831,
+      "loss": 3.5857,
+      "step": 19700
+    },
+    {
+      "epoch": 0.198,
+      "grad_norm": 0.11678298562765121,
+      "learning_rate": 0.0003178208630818641,
+      "loss": 3.5852,
+      "step": 19800
+    },
+    {
+      "epoch": 0.199,
+      "grad_norm": 0.10398001223802567,
+      "learning_rate": 0.00031702131247412076,
+      "loss": 3.5837,
+      "step": 19900
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.09851796180009842,
+      "learning_rate": 0.00031622776601683794,
+      "loss": 3.5844,
+      "step": 20000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 4.0078125,
+      "eval_runtime": 33.1319,
+      "eval_samples_per_second": 18175.498,
+      "eval_steps_per_second": 17.777,
+      "step": 20000
+    },
+    {
+      "epoch": 0.201,
+      "grad_norm": 0.11415958404541016,
+      "learning_rate": 0.00031544014893825583,
+      "loss": 3.5827,
+      "step": 20100
+    },
+    {
+      "epoch": 0.202,
+      "grad_norm": 0.14236243069171906,
+      "learning_rate": 0.0003146583877637763,
+      "loss": 3.5815,
+      "step": 20200
+    },
+    {
+      "epoch": 0.203,
+      "grad_norm": 0.10967651009559631,
+      "learning_rate": 0.0003138824102871722,
+      "loss": 3.5822,
+      "step": 20300
+    },
+    {
+      "epoch": 0.204,
+      "grad_norm": 0.0956553965806961,
+      "learning_rate": 0.0003131121455425748,
+      "loss": 3.5814,
+      "step": 20400
+    },
+    {
+      "epoch": 0.205,
+      "grad_norm": 0.10149464756250381,
+      "learning_rate": 0.0003123475237772121,
+      "loss": 3.5802,
+      "step": 20500
+    },
+    {
+      "epoch": 0.206,
+      "grad_norm": 0.12360329180955887,
+      "learning_rate": 0.0003115884764248779,
+      "loss": 3.5806,
+      "step": 20600
+    },
+    {
+      "epoch": 0.207,
+      "grad_norm": 0.09695049375295639,
+      "learning_rate": 0.0003108349360801046,
+      "loss": 3.5792,
+      "step": 20700
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.09988222271203995,
+      "learning_rate": 0.0003100868364730211,
+      "loss": 3.5786,
+      "step": 20800
+    },
+    {
+      "epoch": 0.209,
+      "grad_norm": 0.10975050926208496,
+      "learning_rate": 0.000309344112444873,
+      "loss": 3.5775,
+      "step": 20900
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.11329706013202667,
+      "learning_rate": 0.00030860669992418383,
+      "loss": 3.579,
+      "step": 21000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 3.99609375,
+      "eval_runtime": 32.9688,
+      "eval_samples_per_second": 18265.41,
+      "eval_steps_per_second": 17.865,
+      "step": 21000
+    },
+    {
+      "epoch": 0.211,
+      "grad_norm": 0.10806494951248169,
+      "learning_rate": 0.00030787453590353956,
+      "loss": 3.5766,
+      "step": 21100
+    },
+    {
+      "epoch": 0.212,
+      "grad_norm": 0.10181587189435959,
+      "learning_rate": 0.0003071475584169756,
+      "loss": 3.578,
+      "step": 21200
+    },
+    {
+      "epoch": 0.213,
+      "grad_norm": 0.1141391471028328,
+      "learning_rate": 0.00030642570651794775,
+      "loss": 3.5747,
+      "step": 21300
+    },
+    {
+      "epoch": 0.214,
+      "grad_norm": 0.0977480486035347,
+      "learning_rate": 0.00030570892025787156,
+      "loss": 3.5767,
+      "step": 21400
+    },
+    {
+      "epoch": 0.215,
+      "grad_norm": 0.12418569624423981,
+      "learning_rate": 0.00030499714066520935,
+      "loss": 3.5739,
+      "step": 21500
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 0.10716783255338669,
+      "learning_rate": 0.00030429030972509223,
+      "loss": 3.5754,
+      "step": 21600
+    },
+    {
+      "epoch": 0.217,
+      "grad_norm": 0.11646725982427597,
+      "learning_rate": 0.0003035883703594582,
+      "loss": 3.5718,
+      "step": 21700
+    },
+    {
+      "epoch": 0.218,
+      "grad_norm": 0.10092218220233917,
+      "learning_rate": 0.00030289126640769133,
+      "loss": 3.575,
+      "step": 21800
+    },
+    {
+      "epoch": 0.219,
+      "grad_norm": 0.10740833729505539,
+      "learning_rate": 0.0003021989426077497,
+      "loss": 3.5721,
+      "step": 21900
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11317898333072662,
+      "learning_rate": 0.00030151134457776364,
+      "loss": 3.5709,
+      "step": 22000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 3.9921875,
+      "eval_runtime": 32.9711,
+      "eval_samples_per_second": 18264.112,
+      "eval_steps_per_second": 17.864,
+      "step": 22000
+    },
+    {
+      "epoch": 0.221,
+      "grad_norm": 0.10360657423734665,
+      "learning_rate": 0.0003008284187980934,
+      "loss": 3.57,
+      "step": 22100
+    },
+    {
+      "epoch": 0.222,
+      "grad_norm": 0.12486978620290756,
+      "learning_rate": 0.0003001501125938321,
+      "loss": 3.5688,
+      "step": 22200
+    },
+    {
+      "epoch": 0.223,
+      "grad_norm": 0.09880336374044418,
+      "learning_rate": 0.00029947637411773995,
+      "loss": 3.5679,
+      "step": 22300
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.116917684674263,
+      "learning_rate": 0.00029880715233359837,
+      "loss": 3.5674,
+      "step": 22400
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 0.10839434713125229,
+      "learning_rate": 0.00029814239699997195,
+      "loss": 3.5662,
+      "step": 22500
+    },
+    {
+      "epoch": 0.226,
+      "grad_norm": 0.12070871144533157,
+      "learning_rate": 0.0002974820586543648,
+      "loss": 3.5661,
+      "step": 22600
+    },
+    {
+      "epoch": 0.227,
+      "grad_norm": 0.1208093985915184,
+      "learning_rate": 0.0002968260885977624,
+      "loss": 3.5677,
+      "step": 22700
+    },
+    {
+      "epoch": 0.228,
+      "grad_norm": 0.1058797761797905,
+      "learning_rate": 0.00029617443887954616,
+      "loss": 3.5673,
+      "step": 22800
+    },
+    {
+      "epoch": 0.229,
+      "grad_norm": 0.10531262308359146,
+      "learning_rate": 0.00029552706228277086,
+      "loss": 3.5658,
+      "step": 22900
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.09782661497592926,
+      "learning_rate": 0.0002948839123097943,
+      "loss": 3.5664,
+      "step": 23000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 3.98828125,
+      "eval_runtime": 32.9558,
+      "eval_samples_per_second": 18272.61,
+      "eval_steps_per_second": 17.872,
+      "step": 23000
+    },
+    {
+      "epoch": 0.231,
+      "grad_norm": 0.09744828194379807,
+      "learning_rate": 0.00029424494316824986,
+      "loss": 3.5646,
+      "step": 23100
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 0.10593089461326599,
+      "learning_rate": 0.00029361010975735173,
+      "loss": 3.5659,
+      "step": 23200
+    },
+    {
+      "epoch": 0.233,
+      "grad_norm": 0.13277162611484528,
+      "learning_rate": 0.0002929793676545238,
+      "loss": 3.563,
+      "step": 23300
+    },
+    {
+      "epoch": 0.234,
+      "grad_norm": 0.09761735796928406,
+      "learning_rate": 0.0002923526731023431,
+      "loss": 3.5636,
+      "step": 23400
+    },
+    {
+      "epoch": 0.235,
+      "grad_norm": 0.10143497586250305,
+      "learning_rate": 0.0002917299829957891,
+      "loss": 3.5624,
+      "step": 23500
+    },
+    {
+      "epoch": 0.236,
+      "grad_norm": 0.09824883192777634,
+      "learning_rate": 0.000291111254869791,
+      "loss": 3.5625,
+      "step": 23600
+    },
+    {
+      "epoch": 0.237,
+      "grad_norm": 0.1004340797662735,
+      "learning_rate": 0.0002904964468870634,
+      "loss": 3.5599,
+      "step": 23700
+    },
+    {
+      "epoch": 0.238,
+      "grad_norm": 0.10824216157197952,
+      "learning_rate": 0.00028988551782622426,
+      "loss": 3.5618,
+      "step": 23800
+    },
+    {
+      "epoch": 0.239,
+      "grad_norm": 0.09485534578561783,
+      "learning_rate": 0.0002892784270701859,
+      "loss": 3.563,
+      "step": 23900
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.09868928045034409,
+      "learning_rate": 0.00028867513459481295,
+      "loss": 3.5597,
+      "step": 24000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 3.984375,
+      "eval_runtime": 33.116,
+      "eval_samples_per_second": 18184.205,
+      "eval_steps_per_second": 17.786,
+      "step": 24000
+    },
+    {
+      "epoch": 0.241,
+      "grad_norm": 0.10118967294692993,
+      "learning_rate": 0.00028807560095783866,
+      "loss": 3.5607,
+      "step": 24100
+    },
+    {
+      "epoch": 0.242,
+      "grad_norm": 0.11375439912080765,
+      "learning_rate": 0.00028747978728803456,
+      "loss": 3.5614,
+      "step": 24200
+    },
+    {
+      "epoch": 0.243,
+      "grad_norm": 0.1017841324210167,
+      "learning_rate": 0.00028688765527462345,
+      "loss": 3.5578,
+      "step": 24300
+    },
+    {
+      "epoch": 0.244,
+      "grad_norm": 0.10626488924026489,
+      "learning_rate": 0.0002862991671569341,
+      "loss": 3.5591,
+      "step": 24400
+    },
+    {
+      "epoch": 0.245,
+      "grad_norm": 0.12246066331863403,
+      "learning_rate": 0.0002857142857142857,
+      "loss": 3.5587,
+      "step": 24500
+    },
+    {
+      "epoch": 0.246,
+      "grad_norm": 0.11167100071907043,
+      "learning_rate": 0.0002851329742561005,
+      "loss": 3.5573,
+      "step": 24600
+    },
+    {
+      "epoch": 0.247,
+      "grad_norm": 0.12268108874559402,
+      "learning_rate": 0.0002845551966122361,
+      "loss": 3.5569,
+      "step": 24700
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 0.09834783524274826,
+      "learning_rate": 0.0002839809171235324,
+      "loss": 3.5551,
+      "step": 24800
+    },
+    {
+      "epoch": 0.249,
+      "grad_norm": 0.11501438915729523,
+      "learning_rate": 0.00028341010063256787,
+      "loss": 3.5549,
+      "step": 24900
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12867997586727142,
+      "learning_rate": 0.000282842712474619,
+      "loss": 3.5538,
+      "step": 25000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 3.978515625,
+      "eval_runtime": 33.0607,
+      "eval_samples_per_second": 18214.619,
+      "eval_steps_per_second": 17.816,
+      "step": 25000
+    },
+    {
+      "epoch": 0.251,
+      "grad_norm": 0.09644894301891327,
+      "learning_rate": 0.0002822787184688183,
+      "loss": 3.556,
+      "step": 25100
+    },
+    {
+      "epoch": 0.252,
+      "grad_norm": 0.10058575868606567,
+      "learning_rate": 0.0002817180849095055,
+      "loss": 3.5537,
+      "step": 25200
+    },
+    {
+      "epoch": 0.253,
+      "grad_norm": 0.10156785696744919,
+      "learning_rate": 0.0002811607785577666,
+      "loss": 3.5531,
+      "step": 25300
+    },
+    {
+      "epoch": 0.254,
+      "grad_norm": 0.0953749492764473,
+      "learning_rate": 0.0002806067666331569,
+      "loss": 3.5514,
+      "step": 25400
+    },
+    {
+      "epoch": 0.255,
+      "grad_norm": 0.1226545125246048,
+      "learning_rate": 0.00028005601680560193,
+      "loss": 3.5512,
+      "step": 25500
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.09581299126148224,
+      "learning_rate": 0.00027950849718747374,
+      "loss": 3.5515,
+      "step": 25600
+    },
+    {
+      "epoch": 0.257,
+      "grad_norm": 0.12539328634738922,
+      "learning_rate": 0.0002789641763258353,
+      "loss": 3.5513,
+      "step": 25700
+    },
+    {
+      "epoch": 0.258,
+      "grad_norm": 0.12774671614170074,
+      "learning_rate": 0.0002784230231948523,
+      "loss": 3.552,
+      "step": 25800
+    },
+    {
+      "epoch": 0.259,
+      "grad_norm": 0.10204903036355972,
+      "learning_rate": 0.00027788500718836423,
+      "loss": 3.5499,
+      "step": 25900
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.09892131388187408,
+      "learning_rate": 0.0002773500981126146,
+      "loss": 3.551,
+      "step": 26000
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 3.98046875,
+      "eval_runtime": 32.9829,
+      "eval_samples_per_second": 18257.575,
+      "eval_steps_per_second": 17.858,
+      "step": 26000
+    },
+    {
+      "epoch": 0.261,
+      "grad_norm": 0.09041019529104233,
+      "learning_rate": 0.0002768182661791332,
+      "loss": 3.5483,
+      "step": 26100
+    },
+    {
+      "epoch": 0.262,
+      "grad_norm": 0.09553086757659912,
+      "learning_rate": 0.00027628948199776883,
+      "loss": 3.546,
+      "step": 26200
+    },
+    {
+      "epoch": 0.263,
+      "grad_norm": 0.09868323802947998,
+      "learning_rate": 0.00027576371656986686,
+      "loss": 3.5477,
+      "step": 26300
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 0.12876607477664948,
+      "learning_rate": 0.00027524094128159016,
+      "loss": 3.5483,
+      "step": 26400
+    },
+    {
+      "epoch": 0.265,
+      "grad_norm": 0.09975658357143402,
+      "learning_rate": 0.0002747211278973781,
+      "loss": 3.5484,
+      "step": 26500
+    },
+    {
+      "epoch": 0.266,
+      "grad_norm": 0.12821029126644135,
+      "learning_rate": 0.0002742042485535409,
+      "loss": 3.5471,
+      "step": 26600
+    },
+    {
+      "epoch": 0.267,
+      "grad_norm": 0.0922236442565918,
+      "learning_rate": 0.00027369027575198666,
+      "loss": 3.5473,
+      "step": 26700
+    },
+    {
+      "epoch": 0.268,
+      "grad_norm": 0.11472651362419128,
+      "learning_rate": 0.0002731791823540765,
+      "loss": 3.5465,
+      "step": 26800
+    },
+    {
+      "epoch": 0.269,
+      "grad_norm": 0.10681800544261932,
+      "learning_rate": 0.0002726709415746059,
+      "loss": 3.544,
+      "step": 26900
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1048787459731102,
+      "learning_rate": 0.0002721655269759087,
+      "loss": 3.545,
+      "step": 27000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 3.97265625,
+      "eval_runtime": 32.9837,
+      "eval_samples_per_second": 18257.114,
+      "eval_steps_per_second": 17.857,
+      "step": 27000
+    },
+    {
+      "epoch": 0.271,
+      "grad_norm": 0.10347763448953629,
+      "learning_rate": 0.00027166291246208065,
+      "loss": 3.5436,
+      "step": 27100
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.10141801089048386,
+      "learning_rate": 0.0002711630722733202,
+      "loss": 3.5438,
+      "step": 27200
+    },
+    {
+      "epoch": 0.273,
+      "grad_norm": 0.09036429226398468,
+      "learning_rate": 0.00027066598098038336,
+      "loss": 3.543,
+      "step": 27300
+    },
+    {
+      "epoch": 0.274,
+      "grad_norm": 0.09406576305627823,
+      "learning_rate": 0.0002701716134791496,
+      "loss": 3.5427,
+      "step": 27400
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 0.09352946281433105,
+      "learning_rate": 0.00026967994498529687,
+      "loss": 3.5417,
+      "step": 27500
+    },
+    {
+      "epoch": 0.276,
+      "grad_norm": 0.09936648607254028,
+      "learning_rate": 0.00026919095102908273,
+      "loss": 3.5428,
+      "step": 27600
+    },
+    {
+      "epoch": 0.277,
+      "grad_norm": 0.10206812620162964,
+      "learning_rate": 0.00026870460745022953,
+      "loss": 3.5424,
+      "step": 27700
+    },
+    {
+      "epoch": 0.278,
+      "grad_norm": 0.11528582870960236,
+      "learning_rate": 0.00026822089039291,
+      "loss": 3.5432,
+      "step": 27800
+    },
+    {
+      "epoch": 0.279,
+      "grad_norm": 0.09294348955154419,
+      "learning_rate": 0.00026773977630083294,
+      "loss": 3.5409,
+      "step": 27900
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.10319820046424866,
+      "learning_rate": 0.0002672612419124244,
+      "loss": 3.5414,
+      "step": 28000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 3.962890625,
+      "eval_runtime": 32.9656,
+      "eval_samples_per_second": 18267.183,
+      "eval_steps_per_second": 17.867,
+      "step": 28000
+    },
+    {
+      "epoch": 0.281,
+      "grad_norm": 0.12130926549434662,
+      "learning_rate": 0.0002667852642561041,
+      "loss": 3.5419,
+      "step": 28100
+    },
+    {
+      "epoch": 0.282,
+      "grad_norm": 0.1062818095088005,
+      "learning_rate": 0.00026631182064565373,
+      "loss": 3.5414,
+      "step": 28200
+    },
+    {
+      "epoch": 0.283,
+      "grad_norm": 0.10452710092067719,
+      "learning_rate": 0.0002658408886756753,
+      "loss": 3.5429,
+      "step": 28300
+    },
+    {
+      "epoch": 0.284,
+      "grad_norm": 0.15059900283813477,
+      "learning_rate": 0.0002653724462171376,
+      "loss": 3.5387,
+      "step": 28400
+    },
+    {
+      "epoch": 0.285,
+      "grad_norm": 0.1282389611005783,
+      "learning_rate": 0.00026490647141300875,
+      "loss": 3.5371,
+      "step": 28500
+    },
+    {
+      "epoch": 0.286,
+      "grad_norm": 0.09370238333940506,
+      "learning_rate": 0.0002644429426739725,
+      "loss": 3.5379,
+      "step": 28600
+    },
+    {
+      "epoch": 0.287,
+      "grad_norm": 0.11192005127668381,
+      "learning_rate": 0.00026398183867422733,
+      "loss": 3.5394,
+      "step": 28700
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.11204060167074203,
+      "learning_rate": 0.00026352313834736497,
+      "loss": 3.5378,
+      "step": 28800
+    },
+    {
+      "epoch": 0.289,
+      "grad_norm": 0.11505492776632309,
+      "learning_rate": 0.0002630668208823282,
+      "loss": 3.5368,
+      "step": 28900
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.09516780078411102,
+      "learning_rate": 0.0002626128657194451,
+      "loss": 3.5368,
+      "step": 29000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 3.962890625,
+      "eval_runtime": 32.9727,
+      "eval_samples_per_second": 18263.215,
+      "eval_steps_per_second": 17.863,
+      "step": 29000
+    },
+    {
+      "epoch": 0.291,
+      "grad_norm": 0.1098528504371643,
+      "learning_rate": 0.00026216125254653817,
+      "loss": 3.5356,
+      "step": 29100
+    },
+    {
+      "epoch": 0.292,
+      "grad_norm": 0.09026819467544556,
+      "learning_rate": 0.0002617119612951068,
+      "loss": 3.5342,
+      "step": 29200
+    },
+    {
+      "epoch": 0.293,
+      "grad_norm": 0.09625507891178131,
+      "learning_rate": 0.00026126497213658205,
+      "loss": 3.538,
+      "step": 29300
+    },
+    {
+      "epoch": 0.294,
+      "grad_norm": 0.13931472599506378,
+      "learning_rate": 0.00026082026547865055,
+      "loss": 3.5355,
+      "step": 29400
+    },
+    {
+      "epoch": 0.295,
+      "grad_norm": 0.09799457341432571,
+      "learning_rate": 0.0002603778219616478,
+      "loss": 3.5323,
+      "step": 29500
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 0.10035666823387146,
+      "learning_rate": 0.00025993762245501815,
+      "loss": 3.5353,
+      "step": 29600
+    },
+    {
+      "epoch": 0.297,
+      "grad_norm": 0.09533281624317169,
+      "learning_rate": 0.000259499648053841,
+      "loss": 3.5361,
+      "step": 29700
+    },
+    {
+      "epoch": 0.298,
+      "grad_norm": 0.1048363596200943,
+      "learning_rate": 0.00025906388007541984,
+      "loss": 3.533,
+      "step": 29800
+    },
+    {
+      "epoch": 0.299,
+      "grad_norm": 0.1037365198135376,
+      "learning_rate": 0.00025863030005593586,
+      "loss": 3.5328,
+      "step": 29900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.09484585374593735,
+      "learning_rate": 0.0002581988897471611,
+      "loss": 3.5336,
+      "step": 30000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 3.958984375,
+      "eval_runtime": 32.9766,
+      "eval_samples_per_second": 18261.09,
+      "eval_steps_per_second": 17.861,
+      "step": 30000
+    },
+    {
+      "epoch": 0.301,
+      "grad_norm": 0.09425152838230133,
+      "learning_rate": 0.00025776963111323354,
+      "loss": 3.5315,
+      "step": 30100
+    },
+    {
+      "epoch": 0.302,
+      "grad_norm": 0.09814691543579102,
+      "learning_rate": 0.0002573425063274894,
+      "loss": 3.5347,
+      "step": 30200
+    },
+    {
+      "epoch": 0.303,
+      "grad_norm": 0.1039854884147644,
+      "learning_rate": 0.00025691749776935395,
+      "loss": 3.5307,
+      "step": 30300
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.11937417089939117,
+      "learning_rate": 0.0002564945880212886,
+      "loss": 3.5315,
+      "step": 30400
+    },
+    {
+      "epoch": 0.305,
+      "grad_norm": 0.10565177351236343,
+      "learning_rate": 0.000256073759865792,
+      "loss": 3.5316,
+      "step": 30500
+    },
+    {
+      "epoch": 0.306,
+      "grad_norm": 0.12661132216453552,
+      "learning_rate": 0.00025565499628245683,
+      "loss": 3.5289,
+      "step": 30600
+    },
+    {
+      "epoch": 0.307,
+      "grad_norm": 0.10317707061767578,
+      "learning_rate": 0.00025523828044507794,
+      "loss": 3.5307,
+      "step": 30700
+    },
+    {
+      "epoch": 0.308,
+      "grad_norm": 0.10436379164457321,
+      "learning_rate": 0.00025482359571881276,
+      "loss": 3.5295,
+      "step": 30800
+    },
+    {
+      "epoch": 0.309,
+      "grad_norm": 0.09864766150712967,
+      "learning_rate": 0.0002544109256573922,
+      "loss": 3.53,
+      "step": 30900
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.0929613932967186,
+      "learning_rate": 0.000254000254000381,
+      "loss": 3.5272,
+      "step": 31000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 3.955078125,
+      "eval_runtime": 33.0115,
+      "eval_samples_per_second": 18241.739,
+      "eval_steps_per_second": 17.842,
+      "step": 31000
+    },
+    {
+      "epoch": 0.311,
+      "grad_norm": 0.09546711295843124,
+      "learning_rate": 0.00025359156467048686,
+      "loss": 3.528,
+      "step": 31100
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 0.09321217238903046,
+      "learning_rate": 0.00025318484177091664,
+      "loss": 3.5296,
+      "step": 31200
+    },
+    {
+      "epoch": 0.313,
+      "grad_norm": 0.11899935454130173,
+      "learning_rate": 0.00025278006958277935,
+      "loss": 3.528,
+      "step": 31300
+    },
+    {
+      "epoch": 0.314,
+      "grad_norm": 0.09187714010477066,
+      "learning_rate": 0.00025237723256253436,
+      "loss": 3.5265,
+      "step": 31400
+    },
+    {
+      "epoch": 0.315,
+      "grad_norm": 0.09416347742080688,
+      "learning_rate": 0.0002519763153394848,
+      "loss": 3.5277,
+      "step": 31500
+    },
+    {
+      "epoch": 0.316,
+      "grad_norm": 0.09648656100034714,
+      "learning_rate": 0.0002515773027133138,
+      "loss": 3.5246,
+      "step": 31600
+    },
+    {
+      "epoch": 0.317,
+      "grad_norm": 0.09706927835941315,
+      "learning_rate": 0.0002511801796516642,
+      "loss": 3.5266,
+      "step": 31700
+    },
+    {
+      "epoch": 0.318,
+      "grad_norm": 0.10058547556400299,
+      "learning_rate": 0.00025078493128775957,
+      "loss": 3.5249,
+      "step": 31800
+    },
+    {
+      "epoch": 0.319,
+      "grad_norm": 0.11078114062547684,
+      "learning_rate": 0.0002503915429180672,
+      "loss": 3.5258,
+      "step": 31900
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.09327193349599838,
+      "learning_rate": 0.00025,
+      "loss": 3.5257,
+      "step": 32000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 3.953125,
+      "eval_runtime": 32.9821,
+      "eval_samples_per_second": 18258.033,
+      "eval_steps_per_second": 17.858,
+      "step": 32000
+    },
+    {
+      "epoch": 0.321,
+      "grad_norm": 0.10520542412996292,
+      "learning_rate": 0.0002496102881496589,
+      "loss": 3.5264,
+      "step": 32100
+    },
+    {
+      "epoch": 0.322,
+      "grad_norm": 0.11748110502958298,
+      "learning_rate": 0.0002492223931396134,
+      "loss": 3.526,
+      "step": 32200
+    },
+    {
+      "epoch": 0.323,
+      "grad_norm": 0.11243593692779541,
+      "learning_rate": 0.0002488363008967198,
+      "loss": 3.5244,
+      "step": 32300
+    },
+    {
+      "epoch": 0.324,
+      "grad_norm": 0.09914110600948334,
+      "learning_rate": 0.00024845199749997667,
+      "loss": 3.5249,
+      "step": 32400
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 0.0963945984840393,
+      "learning_rate": 0.00024806946917841694,
+      "loss": 3.5244,
+      "step": 32500
+    },
+    {
+      "epoch": 0.326,
+      "grad_norm": 0.09173919260501862,
+      "learning_rate": 0.00024768870230903496,
+      "loss": 3.5236,
+      "step": 32600
+    },
+    {
+      "epoch": 0.327,
+      "grad_norm": 0.1360139101743698,
+      "learning_rate": 0.00024730968341474897,
+      "loss": 3.5213,
+      "step": 32700
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 0.09488269686698914,
+      "learning_rate": 0.0002469323991623974,
+      "loss": 3.5235,
+      "step": 32800
+    },
+    {
+      "epoch": 0.329,
+      "grad_norm": 0.1184873878955841,
+      "learning_rate": 0.00024655683636076893,
+      "loss": 3.5233,
+      "step": 32900
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.09744580835103989,
+      "learning_rate": 0.00024618298195866543,
+      "loss": 3.5208,
+      "step": 33000
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 3.951171875,
+      "eval_runtime": 32.9597,
+      "eval_samples_per_second": 18270.423,
+      "eval_steps_per_second": 17.87,
+      "step": 33000
+    },
+    {
+      "epoch": 0.331,
+      "grad_norm": 0.09271999448537827,
+      "learning_rate": 0.0002458108230429969,
+      "loss": 3.5197,
+      "step": 33100
+    },
+    {
+      "epoch": 0.332,
+      "grad_norm": 0.12717638909816742,
+      "learning_rate": 0.00024544034683690797,
+      "loss": 3.5194,
+      "step": 33200
+    },
+    {
+      "epoch": 0.333,
+      "grad_norm": 0.10852134227752686,
+      "learning_rate": 0.0002450715406979359,
+      "loss": 3.5196,
+      "step": 33300
+    },
+    {
+      "epoch": 0.334,
+      "grad_norm": 0.10992783308029175,
+      "learning_rate": 0.0002447043921161982,
+      "loss": 3.5216,
+      "step": 33400
+    },
+    {
+      "epoch": 0.335,
+      "grad_norm": 0.1337759792804718,
+      "learning_rate": 0.0002443388887126105,
+      "loss": 3.5205,
+      "step": 33500
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.11975846439599991,
+      "learning_rate": 0.00024397501823713327,
+      "loss": 3.5191,
+      "step": 33600
+    },
+    {
+      "epoch": 0.337,
+      "grad_norm": 0.14458173513412476,
+      "learning_rate": 0.00024361276856704796,
+      "loss": 3.5203,
+      "step": 33700
+    },
+    {
+      "epoch": 0.338,
+      "grad_norm": 0.09570927172899246,
+      "learning_rate": 0.00024325212770525995,
+      "loss": 3.5184,
+      "step": 33800
+    },
+    {
+      "epoch": 0.339,
+      "grad_norm": 0.11703662574291229,
+      "learning_rate": 0.00024289308377863016,
+      "loss": 3.5176,
+      "step": 33900
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.09540607035160065,
+      "learning_rate": 0.00024253562503633296,
+      "loss": 3.5168,
+      "step": 34000
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 3.943359375,
+      "eval_runtime": 32.983,
+      "eval_samples_per_second": 18257.51,
+      "eval_steps_per_second": 17.858,
+      "step": 34000
+    },
+    {
+      "epoch": 0.341,
+      "grad_norm": 0.09772911667823792,
+      "learning_rate": 0.00024217973984824137,
+      "loss": 3.5173,
+      "step": 34100
+    },
+    {
+      "epoch": 0.342,
+      "grad_norm": 0.10085777193307877,
+      "learning_rate": 0.0002418254167033372,
+      "loss": 3.5179,
+      "step": 34200
+    },
+    {
+      "epoch": 0.343,
+      "grad_norm": 0.10360076278448105,
+      "learning_rate": 0.0002414726442081476,
+      "loss": 3.5171,
+      "step": 34300
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 0.10194404423236847,
+      "learning_rate": 0.0002411214110852061,
+      "loss": 3.5145,
+      "step": 34400
+    },
+    {
+      "epoch": 0.345,
+      "grad_norm": 0.1001388356089592,
+      "learning_rate": 0.0002407717061715384,
+      "loss": 3.5177,
+      "step": 34500
+    },
+    {
+      "epoch": 0.346,
+      "grad_norm": 0.10188458114862442,
+      "learning_rate": 0.00024042351841717252,
+      "loss": 3.5156,
+      "step": 34600
+    },
+    {
+      "epoch": 0.347,
+      "grad_norm": 0.10750128328800201,
+      "learning_rate": 0.00024007683688367183,
+      "loss": 3.5192,
+      "step": 34700
+    },
+    {
+      "epoch": 0.348,
+      "grad_norm": 0.12118516862392426,
+      "learning_rate": 0.0002397316507426921,
+      "loss": 3.517,
+      "step": 34800
+    },
+    {
+      "epoch": 0.349,
+      "grad_norm": 0.09460828453302383,
+      "learning_rate": 0.0002393879492745607,
+      "loss": 3.5176,
+      "step": 34900
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.10879145562648773,
+      "learning_rate": 0.00023904572186687873,
+      "loss": 3.5135,
+      "step": 35000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 3.94140625,
+      "eval_runtime": 32.9909,
+      "eval_samples_per_second": 18253.178,
+      "eval_steps_per_second": 17.853,
+      "step": 35000
+    },
+    {
+      "epoch": 0.351,
+      "grad_norm": 0.10676248371601105,
+      "learning_rate": 0.00023870495801314433,
+      "loss": 3.5156,
+      "step": 35100
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.09372402727603912,
+      "learning_rate": 0.00023836564731139807,
+      "loss": 3.5155,
+      "step": 35200
+    },
+    {
+      "epoch": 0.353,
+      "grad_norm": 0.09432759135961533,
+      "learning_rate": 0.00023802777946288955,
+      "loss": 3.5148,
+      "step": 35300
+    },
+    {
+      "epoch": 0.354,
+      "grad_norm": 0.13439424335956573,
+      "learning_rate": 0.00023769134427076416,
+      "loss": 3.5149,
+      "step": 35400
+    },
+    {
+      "epoch": 0.355,
+      "grad_norm": 0.09214833378791809,
+      "learning_rate": 0.00023735633163877065,
+      "loss": 3.5141,
+      "step": 35500
+    },
+    {
+      "epoch": 0.356,
+      "grad_norm": 0.08910084515810013,
+      "learning_rate": 0.00023702273156998862,
+      "loss": 3.513,
+      "step": 35600
+    },
+    {
+      "epoch": 0.357,
+      "grad_norm": 0.09308894723653793,
+      "learning_rate": 0.00023669053416557542,
+      "loss": 3.5136,
+      "step": 35700
+    },
+    {
+      "epoch": 0.358,
+      "grad_norm": 0.09210114926099777,
+      "learning_rate": 0.00023635972962353274,
+      "loss": 3.513,
+      "step": 35800
+    },
+    {
+      "epoch": 0.359,
+      "grad_norm": 0.0928545594215393,
+      "learning_rate": 0.0002360303082374915,
+      "loss": 3.5111,
+      "step": 35900
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.09326168894767761,
+      "learning_rate": 0.00023570226039551587,
+      "loss": 3.5124,
+      "step": 36000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 3.9453125,
+      "eval_runtime": 32.9906,
+      "eval_samples_per_second": 18253.307,
+      "eval_steps_per_second": 17.854,
+      "step": 36000
+    },
+    {
+      "epoch": 0.361,
+      "grad_norm": 0.11911537498235703,
+      "learning_rate": 0.00023537557657892522,
+      "loss": 3.5123,
+      "step": 36100
+    },
+    {
+      "epoch": 0.362,
+      "grad_norm": 0.12963728606700897,
+      "learning_rate": 0.00023505024736113423,
+      "loss": 3.5114,
+      "step": 36200
+    },
+    {
+      "epoch": 0.363,
+      "grad_norm": 0.10376954078674316,
+      "learning_rate": 0.00023472626340651013,
+      "loss": 3.5101,
+      "step": 36300
+    },
+    {
+      "epoch": 0.364,
+      "grad_norm": 0.09793874621391296,
+      "learning_rate": 0.00023440361546924772,
+      "loss": 3.5112,
+      "step": 36400
+    },
+    {
+      "epoch": 0.365,
+      "grad_norm": 0.1108843982219696,
+      "learning_rate": 0.00023408229439226115,
+      "loss": 3.509,
+      "step": 36500
+    },
+    {
+      "epoch": 0.366,
+      "grad_norm": 0.17180891335010529,
+      "learning_rate": 0.0002337622911060922,
+      "loss": 3.5084,
+      "step": 36600
+    },
+    {
+      "epoch": 0.367,
+      "grad_norm": 0.12893500924110413,
+      "learning_rate": 0.00023344359662783541,
+      "loss": 3.5105,
+      "step": 36700
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.09860619902610779,
+      "learning_rate": 0.00023312620206007844,
+      "loss": 3.5104,
+      "step": 36800
+    },
+    {
+      "epoch": 0.369,
+      "grad_norm": 0.12415798008441925,
+      "learning_rate": 0.00023281009858985938,
+      "loss": 3.511,
+      "step": 36900
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.08730453252792358,
+      "learning_rate": 0.00023249527748763857,
+      "loss": 3.51,
+      "step": 37000
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 3.94140625,
+      "eval_runtime": 32.9875,
+      "eval_samples_per_second": 18255.033,
+      "eval_steps_per_second": 17.855,
+      "step": 37000
+    },
+    {
+      "epoch": 0.371,
+      "grad_norm": 0.09179120510816574,
+      "learning_rate": 0.000232181730106286,
+      "loss": 3.5092,
+      "step": 37100
+    },
+    {
+      "epoch": 0.372,
+      "grad_norm": 0.09930966049432755,
+      "learning_rate": 0.00023186944788008412,
+      "loss": 3.5109,
+      "step": 37200
+    },
+    {
+      "epoch": 0.373,
+      "grad_norm": 0.09958541393280029,
+      "learning_rate": 0.00023155842232374464,
+      "loss": 3.5109,
+      "step": 37300
+    },
+    {
+      "epoch": 0.374,
+      "grad_norm": 0.12271804362535477,
+      "learning_rate": 0.00023124864503144013,
+      "loss": 3.5105,
+      "step": 37400
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.09509623795747757,
+      "learning_rate": 0.00023094010767585029,
+      "loss": 3.5071,
+      "step": 37500
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 0.10307575762271881,
+      "learning_rate": 0.00023063280200722128,
+      "loss": 3.509,
+      "step": 37600
+    },
+    {
+      "epoch": 0.377,
+      "grad_norm": 0.11047662794589996,
+      "learning_rate": 0.00023032671985243937,
+      "loss": 3.507,
+      "step": 37700
+    },
+    {
+      "epoch": 0.378,
+      "grad_norm": 0.1369996964931488,
+      "learning_rate": 0.00023002185311411807,
+      "loss": 3.5077,
+      "step": 37800
+    },
+    {
+      "epoch": 0.379,
+      "grad_norm": 0.09368377178907394,
+      "learning_rate": 0.0002297181937696983,
+      "loss": 3.5079,
+      "step": 37900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.0933164581656456,
+      "learning_rate": 0.00022941573387056174,
+      "loss": 3.5061,
+      "step": 38000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 3.94140625,
+      "eval_runtime": 33.0024,
+      "eval_samples_per_second": 18246.79,
+      "eval_steps_per_second": 17.847,
+      "step": 38000
+    },
+    {
+      "epoch": 0.381,
+      "grad_norm": 0.11005818843841553,
+      "learning_rate": 0.0002291144655411569,
+      "loss": 3.5067,
+      "step": 38100
+    },
+    {
+      "epoch": 0.382,
+      "grad_norm": 0.09861373901367188,
+      "learning_rate": 0.00022881438097813774,
+      "loss": 3.5034,
+      "step": 38200
+    },
+    {
+      "epoch": 0.383,
+      "grad_norm": 0.10625416785478592,
+      "learning_rate": 0.00022851547244951505,
+      "loss": 3.5034,
+      "step": 38300
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.09815394133329391,
+      "learning_rate": 0.00022821773229381924,
+      "loss": 3.505,
+      "step": 38400
+    },
+    {
+      "epoch": 0.385,
+      "grad_norm": 0.11951077729463577,
+      "learning_rate": 0.0002279211529192759,
+      "loss": 3.5052,
+      "step": 38500
+    },
+    {
+      "epoch": 0.386,
+      "grad_norm": 0.09885293990373611,
+      "learning_rate": 0.0002276257268029927,
+      "loss": 3.5032,
+      "step": 38600
+    },
+    {
+      "epoch": 0.387,
+      "grad_norm": 0.10494934767484665,
+      "learning_rate": 0.0002273314464901578,
+      "loss": 3.5018,
+      "step": 38700
+    },
+    {
+      "epoch": 0.388,
+      "grad_norm": 0.1292721927165985,
+      "learning_rate": 0.0002270383045932499,
+      "loss": 3.5039,
+      "step": 38800
+    },
+    {
+      "epoch": 0.389,
+      "grad_norm": 0.0997849851846695,
+      "learning_rate": 0.00022674629379125912,
+      "loss": 3.5016,
+      "step": 38900
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1319294422864914,
+      "learning_rate": 0.00022645540682891913,
+      "loss": 3.5044,
+      "step": 39000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 3.931640625,
+      "eval_runtime": 32.9931,
+      "eval_samples_per_second": 18251.944,
+      "eval_steps_per_second": 17.852,
+      "step": 39000
+    },
+    {
+      "epoch": 0.391,
+      "grad_norm": 0.11860304325819016,
+      "learning_rate": 0.0002261656365159503,
+      "loss": 3.5038,
+      "step": 39100
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 0.1081485003232956,
+      "learning_rate": 0.00022587697572631283,
+      "loss": 3.5016,
+      "step": 39200
+    },
+    {
+      "epoch": 0.393,
+      "grad_norm": 0.10177210718393326,
+      "learning_rate": 0.00022558941739747075,
+      "loss": 3.5034,
+      "step": 39300
+    },
+    {
+      "epoch": 0.394,
+      "grad_norm": 0.08930639922618866,
+      "learning_rate": 0.00022530295452966644,
+      "loss": 3.5026,
+      "step": 39400
+    },
+    {
+      "epoch": 0.395,
+      "grad_norm": 0.10102177411317825,
+      "learning_rate": 0.0002250175801852048,
+      "loss": 3.5022,
+      "step": 39500
+    },
+    {
+      "epoch": 0.396,
+      "grad_norm": 0.11118755489587784,
+      "learning_rate": 0.00022473328748774736,
+      "loss": 3.5019,
+      "step": 39600
+    },
+    {
+      "epoch": 0.397,
+      "grad_norm": 0.09080816060304642,
+      "learning_rate": 0.00022445006962161677,
+      "loss": 3.5018,
+      "step": 39700
+    },
+    {
+      "epoch": 0.398,
+      "grad_norm": 0.09106365591287613,
+      "learning_rate": 0.00022416791983111015,
+      "loss": 3.4995,
+      "step": 39800
+    },
+    {
+      "epoch": 0.399,
+      "grad_norm": 0.08732519298791885,
+      "learning_rate": 0.0002238868314198225,
+      "loss": 3.4991,
+      "step": 39900
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.08918973803520203,
+      "learning_rate": 0.00022360679774997898,
+      "loss": 3.501,
+      "step": 40000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 3.931640625,
+      "eval_runtime": 33.0068,
+      "eval_samples_per_second": 18244.355,
+      "eval_steps_per_second": 17.845,
+      "step": 40000
+    },
+    {
+      "epoch": 0.401,
+      "grad_norm": 0.08663675934076309,
+      "learning_rate": 0.00022332781224177668,
+      "loss": 3.501,
+      "step": 40100
+    },
+    {
+      "epoch": 0.402,
+      "grad_norm": 0.09693952649831772,
+      "learning_rate": 0.00022304986837273525,
+      "loss": 3.5006,
+      "step": 40200
+    },
+    {
+      "epoch": 0.403,
+      "grad_norm": 0.11651097983121872,
+      "learning_rate": 0.00022277295967705653,
+      "loss": 3.4992,
+      "step": 40300
+    },
+    {
+      "epoch": 0.404,
+      "grad_norm": 0.08836641907691956,
+      "learning_rate": 0.0002224970797449924,
+      "loss": 3.4997,
+      "step": 40400
+    },
+    {
+      "epoch": 0.405,
+      "grad_norm": 0.10767354816198349,
+      "learning_rate": 0.0002222222222222222,
+      "loss": 3.4995,
+      "step": 40500
+    },
+    {
+      "epoch": 0.406,
+      "grad_norm": 0.09033215790987015,
+      "learning_rate": 0.00022194838080923766,
+      "loss": 3.4976,
+      "step": 40600
+    },
+    {
+      "epoch": 0.407,
+      "grad_norm": 0.0916823074221611,
+      "learning_rate": 0.00022167554926073632,
+      "loss": 3.4983,
+      "step": 40700
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 0.13178223371505737,
+      "learning_rate": 0.00022140372138502385,
+      "loss": 3.4957,
+      "step": 40800
+    },
+    {
+      "epoch": 0.409,
+      "grad_norm": 0.1049867570400238,
+      "learning_rate": 0.00022113289104342323,
+      "loss": 3.4973,
+      "step": 40900
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.09196602553129196,
+      "learning_rate": 0.00022086305214969308,
+      "loss": 3.4996,
+      "step": 41000
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 3.9296875,
+      "eval_runtime": 32.9946,
+      "eval_samples_per_second": 18251.123,
+      "eval_steps_per_second": 17.851,
+      "step": 41000
+    },
+    {
+      "epoch": 0.411,
+      "grad_norm": 0.0933445394039154,
+      "learning_rate": 0.00022059419866945278,
+      "loss": 3.4972,
+      "step": 41100
+    },
+    {
+      "epoch": 0.412,
+      "grad_norm": 0.10621432214975357,
+      "learning_rate": 0.00022032632461961585,
+      "loss": 3.4988,
+      "step": 41200
+    },
+    {
+      "epoch": 0.413,
+      "grad_norm": 0.1379343420267105,
+      "learning_rate": 0.00022005942406783078,
+      "loss": 3.4983,
+      "step": 41300
+    },
+    {
+      "epoch": 0.414,
+      "grad_norm": 0.1172490119934082,
+      "learning_rate": 0.00021979349113192902,
+      "loss": 3.4954,
+      "step": 41400
+    },
+    {
+      "epoch": 0.415,
+      "grad_norm": 0.10637906193733215,
+      "learning_rate": 0.0002195285199793807,
+      "loss": 3.4974,
+      "step": 41500
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.10279926657676697,
+      "learning_rate": 0.0002192645048267573,
+      "loss": 3.495,
+      "step": 41600
+    },
+    {
+      "epoch": 0.417,
+      "grad_norm": 0.13635499775409698,
+      "learning_rate": 0.00021900143993920144,
+      "loss": 3.4997,
+      "step": 41700
+    },
+    {
+      "epoch": 0.418,
+      "grad_norm": 0.11549071967601776,
+      "learning_rate": 0.00021873931962990357,
+      "loss": 3.4963,
+      "step": 41800
+    },
+    {
+      "epoch": 0.419,
+      "grad_norm": 0.11906278878450394,
+      "learning_rate": 0.00021847813825958586,
+      "loss": 3.4937,
+      "step": 41900
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.08883855491876602,
+      "learning_rate": 0.0002182178902359924,
+      "loss": 3.4957,
+      "step": 42000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 3.9296875,
+      "eval_runtime": 32.9839,
+      "eval_samples_per_second": 18257.009,
+      "eval_steps_per_second": 17.857,
+      "step": 42000
+    },
+    {
+      "epoch": 0.421,
+      "grad_norm": 0.11999965459108353,
+      "learning_rate": 0.00021795857001338643,
+      "loss": 3.4951,
+      "step": 42100
+    },
+    {
+      "epoch": 0.422,
+      "grad_norm": 0.09602005034685135,
+      "learning_rate": 0.00021770017209205407,
+      "loss": 3.4948,
+      "step": 42200
+    },
+    {
+      "epoch": 0.423,
+      "grad_norm": 0.12526479363441467,
+      "learning_rate": 0.00021744269101781405,
+      "loss": 3.4963,
+      "step": 42300
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 0.08893965929746628,
+      "learning_rate": 0.0002171861213815347,
+      "loss": 3.4941,
+      "step": 42400
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 0.10459540784358978,
+      "learning_rate": 0.00021693045781865617,
+      "loss": 3.4968,
+      "step": 42500
+    },
+    {
+      "epoch": 0.426,
+      "grad_norm": 0.10754283517599106,
+      "learning_rate": 0.00021667569500871976,
+      "loss": 3.4934,
+      "step": 42600
+    },
+    {
+      "epoch": 0.427,
+      "grad_norm": 0.1145748421549797,
+      "learning_rate": 0.0002164218276749025,
+      "loss": 3.4972,
+      "step": 42700
+    },
+    {
+      "epoch": 0.428,
+      "grad_norm": 0.10786953568458557,
+      "learning_rate": 0.00021616885058355848,
+      "loss": 3.4929,
+      "step": 42800
+    },
+    {
+      "epoch": 0.429,
+      "grad_norm": 0.11820059269666672,
+      "learning_rate": 0.00021591675854376523,
+      "loss": 3.4935,
+      "step": 42900
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.08727939426898956,
+      "learning_rate": 0.00021566554640687683,
+      "loss": 3.4932,
+      "step": 43000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 3.92578125,
+      "eval_runtime": 32.9903,
+      "eval_samples_per_second": 18253.464,
+      "eval_steps_per_second": 17.854,
+      "step": 43000
+    },
+    {
+      "epoch": 0.431,
+      "grad_norm": 0.09863859415054321,
+      "learning_rate": 0.00021541520906608185,
+      "loss": 3.4939,
+      "step": 43100
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.10082012414932251,
+      "learning_rate": 0.0002151657414559676,
+      "loss": 3.4948,
+      "step": 43200
+    },
+    {
+      "epoch": 0.433,
+      "grad_norm": 0.15581797063350677,
+      "learning_rate": 0.00021491713855208944,
+      "loss": 3.4922,
+      "step": 43300
+    },
+    {
+      "epoch": 0.434,
+      "grad_norm": 0.08548469096422195,
+      "learning_rate": 0.00021466939537054594,
+      "loss": 3.4897,
+      "step": 43400
+    },
+    {
+      "epoch": 0.435,
+      "grad_norm": 0.09384721517562866,
+      "learning_rate": 0.00021442250696755897,
+      "loss": 3.4923,
+      "step": 43500
+    },
+    {
+      "epoch": 0.436,
+      "grad_norm": 0.11810631304979324,
+      "learning_rate": 0.00021417646843905965,
+      "loss": 3.4907,
+      "step": 43600
+    },
+    {
+      "epoch": 0.437,
+      "grad_norm": 0.10735498368740082,
+      "learning_rate": 0.00021393127492027903,
+      "loss": 3.4928,
+      "step": 43700
+    },
+    {
+      "epoch": 0.438,
+      "grad_norm": 0.09564588963985443,
+      "learning_rate": 0.00021368692158534413,
+      "loss": 3.491,
+      "step": 43800
+    },
+    {
+      "epoch": 0.439,
+      "grad_norm": 0.08804554492235184,
+      "learning_rate": 0.00021344340364687887,
+      "loss": 3.4908,
+      "step": 43900
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.09846536815166473,
+      "learning_rate": 0.00021320071635561042,
+      "loss": 3.4934,
+      "step": 44000
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 3.919921875,
+      "eval_runtime": 33.0158,
+      "eval_samples_per_second": 18239.397,
+      "eval_steps_per_second": 17.84,
+      "step": 44000
+    },
+    {
+      "epoch": 0.441,
+      "grad_norm": 0.09984950721263885,
+      "learning_rate": 0.00021295885499997996,
+      "loss": 3.4919,
+      "step": 44100
+    },
+    {
+      "epoch": 0.442,
+      "grad_norm": 0.09016887098550797,
+      "learning_rate": 0.00021271781490575853,
+      "loss": 3.4899,
+      "step": 44200
+    },
+    {
+      "epoch": 0.443,
+      "grad_norm": 0.08681657910346985,
+      "learning_rate": 0.00021247759143566762,
+      "loss": 3.4906,
+      "step": 44300
+    },
+    {
+      "epoch": 0.444,
+      "grad_norm": 0.09243223071098328,
+      "learning_rate": 0.00021223817998900448,
+      "loss": 3.4897,
+      "step": 44400
+    },
+    {
+      "epoch": 0.445,
+      "grad_norm": 0.08925904333591461,
+      "learning_rate": 0.000211999576001272,
+      "loss": 3.4917,
+      "step": 44500
+    },
+    {
+      "epoch": 0.446,
+      "grad_norm": 0.09652265906333923,
+      "learning_rate": 0.0002117617749438134,
+      "loss": 3.4887,
+      "step": 44600
+    },
+    {
+      "epoch": 0.447,
+      "grad_norm": 0.087586909532547,
+      "learning_rate": 0.0002115247723234508,
+      "loss": 3.4895,
+      "step": 44700
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.11652135103940964,
+      "learning_rate": 0.00021128856368212914,
+      "loss": 3.4892,
+      "step": 44800
+    },
+    {
+      "epoch": 0.449,
+      "grad_norm": 0.10191888362169266,
+      "learning_rate": 0.00021105314459656365,
+      "loss": 3.4888,
+      "step": 44900
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.09599770605564117,
+      "learning_rate": 0.00021081851067789197,
+      "loss": 3.4892,
+      "step": 45000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 3.91796875,
+      "eval_runtime": 33.1684,
+      "eval_samples_per_second": 18155.468,
+      "eval_steps_per_second": 17.758,
+      "step": 45000
+    },
+    {
+      "epoch": 0.451,
+      "grad_norm": 0.11355946213006973,
+      "learning_rate": 0.00021058465757133065,
+      "loss": 3.4885,
+      "step": 45100
+    },
+    {
+      "epoch": 0.452,
+      "grad_norm": 0.10500317811965942,
+      "learning_rate": 0.0002103515809558356,
+      "loss": 3.4864,
+      "step": 45200
+    },
+    {
+      "epoch": 0.453,
+      "grad_norm": 0.12674091756343842,
+      "learning_rate": 0.00021011927654376682,
+      "loss": 3.4877,
+      "step": 45300
+    },
+    {
+      "epoch": 0.454,
+      "grad_norm": 0.09490884095430374,
+      "learning_rate": 0.00020988774008055678,
+      "loss": 3.4868,
+      "step": 45400
+    },
+    {
+      "epoch": 0.455,
+      "grad_norm": 0.0876164361834526,
+      "learning_rate": 0.00020965696734438367,
+      "loss": 3.4869,
+      "step": 45500
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 0.1012013703584671,
+      "learning_rate": 0.00020942695414584775,
+      "loss": 3.4887,
+      "step": 45600
+    },
+    {
+      "epoch": 0.457,
+      "grad_norm": 0.09138456732034683,
+      "learning_rate": 0.00020919769632765198,
+      "loss": 3.4879,
+      "step": 45700
+    },
+    {
+      "epoch": 0.458,
+      "grad_norm": 0.09227514266967773,
+      "learning_rate": 0.00020896918976428644,
+      "loss": 3.4866,
+      "step": 45800
+    },
+    {
+      "epoch": 0.459,
+      "grad_norm": 0.11846242845058441,
+      "learning_rate": 0.0002087414303617165,
+      "loss": 3.4861,
+      "step": 45900
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.09923862665891647,
+      "learning_rate": 0.00020851441405707476,
+      "loss": 3.4869,
+      "step": 46000
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 3.92578125,
+      "eval_runtime": 33.1689,
+      "eval_samples_per_second": 18155.218,
+      "eval_steps_per_second": 17.758,
+      "step": 46000
+    },
+    {
+      "epoch": 0.461,
+      "grad_norm": 0.11270363628864288,
+      "learning_rate": 0.00020828813681835672,
+      "loss": 3.4865,
+      "step": 46100
+    },
+    {
+      "epoch": 0.462,
+      "grad_norm": 0.09331326186656952,
+      "learning_rate": 0.00020806259464411974,
+      "loss": 3.4853,
+      "step": 46200
+    },
+    {
+      "epoch": 0.463,
+      "grad_norm": 0.09471999108791351,
+      "learning_rate": 0.00020783778356318634,
+      "loss": 3.4855,
+      "step": 46300
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.10036460310220718,
+      "learning_rate": 0.00020761369963434993,
+      "loss": 3.486,
+      "step": 46400
+    },
+    {
+      "epoch": 0.465,
+      "grad_norm": 0.09839966893196106,
+      "learning_rate": 0.00020739033894608504,
+      "loss": 3.4836,
+      "step": 46500
+    },
+    {
+      "epoch": 0.466,
+      "grad_norm": 0.13888099789619446,
+      "learning_rate": 0.00020716769761626043,
+      "loss": 3.4845,
+      "step": 46600
+    },
+    {
+      "epoch": 0.467,
+      "grad_norm": 0.11780879646539688,
+      "learning_rate": 0.00020694577179185557,
+      "loss": 3.484,
+      "step": 46700
+    },
+    {
+      "epoch": 0.468,
+      "grad_norm": 0.10853921622037888,
+      "learning_rate": 0.0002067245576486808,
+      "loss": 3.4852,
+      "step": 46800
+    },
+    {
+      "epoch": 0.469,
+      "grad_norm": 0.10998079925775528,
+      "learning_rate": 0.00020650405139110026,
+      "loss": 3.483,
+      "step": 46900
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.09292669594287872,
+      "learning_rate": 0.00020628424925175868,
+      "loss": 3.4847,
+      "step": 47000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 3.912109375,
+      "eval_runtime": 33.15,
+      "eval_samples_per_second": 18165.566,
+      "eval_steps_per_second": 17.768,
+      "step": 47000
+    },
+    {
+      "epoch": 0.471,
+      "grad_norm": 0.09031188488006592,
+      "learning_rate": 0.0002060651474913109,
+      "loss": 3.4842,
+      "step": 47100
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 0.08594627678394318,
+      "learning_rate": 0.00020584674239815457,
+      "loss": 3.4835,
+      "step": 47200
+    },
+    {
+      "epoch": 0.473,
+      "grad_norm": 0.09004069864749908,
+      "learning_rate": 0.00020562903028816627,
+      "loss": 3.4846,
+      "step": 47300
+    },
+    {
+      "epoch": 0.474,
+      "grad_norm": 0.10319007933139801,
+      "learning_rate": 0.00020541200750444027,
+      "loss": 3.4859,
+      "step": 47400
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 0.08985935151576996,
+      "learning_rate": 0.00020519567041703084,
+      "loss": 3.4808,
+      "step": 47500
+    },
+    {
+      "epoch": 0.476,
+      "grad_norm": 0.09091832488775253,
+      "learning_rate": 0.00020498001542269695,
+      "loss": 3.4846,
+      "step": 47600
+    },
+    {
+      "epoch": 0.477,
+      "grad_norm": 0.10014614462852478,
+      "learning_rate": 0.00020476503894465038,
+      "loss": 3.4818,
+      "step": 47700
+    },
+    {
+      "epoch": 0.478,
+      "grad_norm": 0.11294897645711899,
+      "learning_rate": 0.00020455073743230656,
+      "loss": 3.4842,
+      "step": 47800
+    },
+    {
+      "epoch": 0.479,
+      "grad_norm": 0.10061386972665787,
+      "learning_rate": 0.0002043371073610381,
+      "loss": 3.4825,
+      "step": 47900
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.09785416722297668,
+      "learning_rate": 0.00020412414523193154,
+      "loss": 3.4838,
+      "step": 48000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 3.916015625,
+      "eval_runtime": 32.9883,
+      "eval_samples_per_second": 18254.585,
+      "eval_steps_per_second": 17.855,
+      "step": 48000
+    },
+    {
+      "epoch": 0.481,
+      "grad_norm": 0.09520223736763,
+      "learning_rate": 0.0002039118475715464,
+      "loss": 3.485,
+      "step": 48100
+    },
+    {
+      "epoch": 0.482,
+      "grad_norm": 0.0920218825340271,
+      "learning_rate": 0.00020370021093167763,
+      "loss": 3.4828,
+      "step": 48200
+    },
+    {
+      "epoch": 0.483,
+      "grad_norm": 0.12049808353185654,
+      "learning_rate": 0.00020348923188911992,
+      "loss": 3.4809,
+      "step": 48300
+    },
+    {
+      "epoch": 0.484,
+      "grad_norm": 0.11325296759605408,
+      "learning_rate": 0.00020327890704543544,
+      "loss": 3.4795,
+      "step": 48400
+    },
+    {
+      "epoch": 0.485,
+      "grad_norm": 0.0982460081577301,
+      "learning_rate": 0.00020306923302672384,
+      "loss": 3.4791,
+      "step": 48500
+    },
+    {
+      "epoch": 0.486,
+      "grad_norm": 0.10175520926713943,
+      "learning_rate": 0.00020286020648339485,
+      "loss": 3.4821,
+      "step": 48600
+    },
+    {
+      "epoch": 0.487,
+      "grad_norm": 0.10392586141824722,
+      "learning_rate": 0.00020265182408994378,
+      "loss": 3.4809,
+      "step": 48700
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 0.08581174165010452,
+      "learning_rate": 0.000202444082544729,
+      "loss": 3.4805,
+      "step": 48800
+    },
+    {
+      "epoch": 0.489,
+      "grad_norm": 0.0906348004937172,
+      "learning_rate": 0.0002022369785697524,
+      "loss": 3.4827,
+      "step": 48900
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.11659245938062668,
+      "learning_rate": 0.00020203050891044215,
+      "loss": 3.4832,
+      "step": 49000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 3.912109375,
+      "eval_runtime": 33.0,
+      "eval_samples_per_second": 18248.129,
+      "eval_steps_per_second": 17.848,
+      "step": 49000
+    },
+    {
+      "epoch": 0.491,
+      "grad_norm": 0.0962797999382019,
+      "learning_rate": 0.00020182467033543782,
+      "loss": 3.4812,
+      "step": 49100
+    },
+    {
+      "epoch": 0.492,
+      "grad_norm": 0.1375749707221985,
+      "learning_rate": 0.00020161945963637794,
+      "loss": 3.4795,
+      "step": 49200
+    },
+    {
+      "epoch": 0.493,
+      "grad_norm": 0.10032547265291214,
+      "learning_rate": 0.0002014148736276902,
+      "loss": 3.4767,
+      "step": 49300
+    },
+    {
+      "epoch": 0.494,
+      "grad_norm": 0.11045530438423157,
+      "learning_rate": 0.00020121090914638343,
+      "loss": 3.4803,
+      "step": 49400
+    },
+    {
+      "epoch": 0.495,
+      "grad_norm": 0.10294586420059204,
+      "learning_rate": 0.00020100756305184244,
+      "loss": 3.4789,
+      "step": 49500
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.09476403146982193,
+      "learning_rate": 0.00020080483222562474,
+      "loss": 3.4796,
+      "step": 49600
+    },
+    {
+      "epoch": 0.497,
+      "grad_norm": 0.11336295306682587,
+      "learning_rate": 0.00020060271357125986,
+      "loss": 3.4812,
+      "step": 49700
+    },
+    {
+      "epoch": 0.498,
+      "grad_norm": 0.11719685047864914,
+      "learning_rate": 0.0002004012040140506,
+      "loss": 3.4772,
+      "step": 49800
+    },
+    {
+      "epoch": 0.499,
+      "grad_norm": 0.08824979513883591,
+      "learning_rate": 0.00020020030050087656,
+      "loss": 3.4779,
+      "step": 49900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09348143637180328,
+      "learning_rate": 0.0002,
+      "loss": 3.4794,
+      "step": 50000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 3.90625,
+      "eval_runtime": 33.1492,
+      "eval_samples_per_second": 18165.994,
+      "eval_steps_per_second": 17.768,
+      "step": 50000
+    },
+    {
+      "epoch": 0.501,
+      "grad_norm": 0.10792515426874161,
+      "learning_rate": 0.00019980029950087343,
+      "loss": 3.4794,
+      "step": 50100
+    },
+    {
+      "epoch": 0.502,
+      "grad_norm": 0.09229454398155212,
+      "learning_rate": 0.0001996011960139498,
+      "loss": 3.4782,
+      "step": 50200
+    },
+    {
+      "epoch": 0.503,
+      "grad_norm": 0.09463869780302048,
+      "learning_rate": 0.00019940268657049439,
+      "loss": 3.4774,
+      "step": 50300
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 0.09181378781795502,
+      "learning_rate": 0.00019920476822239894,
+      "loss": 3.4771,
+      "step": 50400
+    },
+    {
+      "epoch": 0.505,
+      "grad_norm": 0.09562738984823227,
+      "learning_rate": 0.00019900743804199784,
+      "loss": 3.4781,
+      "step": 50500
+    },
+    {
+      "epoch": 0.506,
+      "grad_norm": 0.08760121464729309,
+      "learning_rate": 0.000198810693121886,
+      "loss": 3.478,
+      "step": 50600
+    },
+    {
+      "epoch": 0.507,
+      "grad_norm": 0.09761682897806168,
+      "learning_rate": 0.00019861453057473934,
+      "loss": 3.4747,
+      "step": 50700
+    },
+    {
+      "epoch": 0.508,
+      "grad_norm": 0.09359412640333176,
+      "learning_rate": 0.00019841894753313627,
+      "loss": 3.4773,
+      "step": 50800
+    },
+    {
+      "epoch": 0.509,
+      "grad_norm": 0.11783072352409363,
+      "learning_rate": 0.00019822394114938217,
+      "loss": 3.4758,
+      "step": 50900
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.09198185056447983,
+      "learning_rate": 0.00019802950859533488,
+      "loss": 3.4783,
+      "step": 51000
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 3.912109375,
+      "eval_runtime": 32.9883,
+      "eval_samples_per_second": 18254.62,
+      "eval_steps_per_second": 17.855,
+      "step": 51000
+    },
+    {
+      "epoch": 0.511,
+      "grad_norm": 0.11417073011398315,
+      "learning_rate": 0.00019783564706223267,
+      "loss": 3.4765,
+      "step": 51100
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.09531737864017487,
+      "learning_rate": 0.00019764235376052369,
+      "loss": 3.477,
+      "step": 51200
+    },
+    {
+      "epoch": 0.513,
+      "grad_norm": 0.11220179498195648,
+      "learning_rate": 0.00019744962591969746,
+      "loss": 3.4757,
+      "step": 51300
+    },
+    {
+      "epoch": 0.514,
+      "grad_norm": 0.09061618894338608,
+      "learning_rate": 0.0001972574607881179,
+      "loss": 3.475,
+      "step": 51400
+    },
+    {
+      "epoch": 0.515,
+      "grad_norm": 0.0985003113746643,
+      "learning_rate": 0.00019706585563285863,
+      "loss": 3.4732,
+      "step": 51500
+    },
+    {
+      "epoch": 0.516,
+      "grad_norm": 0.10033012181520462,
+      "learning_rate": 0.00019687480773953944,
+      "loss": 3.474,
+      "step": 51600
+    },
+    {
+      "epoch": 0.517,
+      "grad_norm": 0.09861289709806442,
+      "learning_rate": 0.00019668431441216495,
+      "loss": 3.473,
+      "step": 51700
+    },
+    {
+      "epoch": 0.518,
+      "grad_norm": 0.13850615918636322,
+      "learning_rate": 0.00019649437297296484,
+      "loss": 3.4756,
+      "step": 51800
+    },
+    {
+      "epoch": 0.519,
+      "grad_norm": 0.12541039288043976,
+      "learning_rate": 0.00019630498076223554,
+      "loss": 3.4747,
+      "step": 51900
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.09361404925584793,
+      "learning_rate": 0.00019611613513818404,
+      "loss": 3.4794,
+      "step": 52000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 3.9140625,
+      "eval_runtime": 32.9814,
+      "eval_samples_per_second": 18258.432,
+      "eval_steps_per_second": 17.859,
+      "step": 52000
+    },
+    {
+      "epoch": 0.521,
+      "grad_norm": 0.09503428637981415,
+      "learning_rate": 0.00019592783347677307,
+      "loss": 3.4736,
+      "step": 52100
+    },
+    {
+      "epoch": 0.522,
+      "grad_norm": 0.10571688413619995,
+      "learning_rate": 0.00019574007317156784,
+      "loss": 3.4746,
+      "step": 52200
+    },
+    {
+      "epoch": 0.523,
+      "grad_norm": 0.12886875867843628,
+      "learning_rate": 0.00019555285163358466,
+      "loss": 3.4716,
+      "step": 52300
+    },
+    {
+      "epoch": 0.524,
+      "grad_norm": 0.11490379273891449,
+      "learning_rate": 0.00019536616629114086,
+      "loss": 3.4764,
+      "step": 52400
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 0.09640925377607346,
+      "learning_rate": 0.00019518001458970665,
+      "loss": 3.4749,
+      "step": 52500
+    },
+    {
+      "epoch": 0.526,
+      "grad_norm": 0.11207705736160278,
+      "learning_rate": 0.00019499439399175798,
+      "loss": 3.4747,
+      "step": 52600
+    },
+    {
+      "epoch": 0.527,
+      "grad_norm": 0.10838747769594193,
+      "learning_rate": 0.00019480930197663147,
+      "loss": 3.4717,
+      "step": 52700
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.11686636507511139,
+      "learning_rate": 0.00019462473604038076,
+      "loss": 3.4725,
+      "step": 52800
+    },
+    {
+      "epoch": 0.529,
+      "grad_norm": 0.0926726907491684,
+      "learning_rate": 0.00019444069369563388,
+      "loss": 3.4735,
+      "step": 52900
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.08765498548746109,
+      "learning_rate": 0.00019425717247145284,
+      "loss": 3.4718,
+      "step": 53000
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 3.912109375,
+      "eval_runtime": 32.9799,
+      "eval_samples_per_second": 18259.264,
+      "eval_steps_per_second": 17.859,
+      "step": 53000
+    },
+    {
+      "epoch": 0.531,
+      "grad_norm": 0.0892767682671547,
+      "learning_rate": 0.000194074169913194,
+      "loss": 3.4701,
+      "step": 53100
+    },
+    {
+      "epoch": 0.532,
+      "grad_norm": 0.10058142244815826,
+      "learning_rate": 0.0001938916835823703,
+      "loss": 3.4715,
+      "step": 53200
+    },
+    {
+      "epoch": 0.533,
+      "grad_norm": 0.09021036326885223,
+      "learning_rate": 0.0001937097110565149,
+      "loss": 3.4736,
+      "step": 53300
+    },
+    {
+      "epoch": 0.534,
+      "grad_norm": 0.10363597422838211,
+      "learning_rate": 0.00019352824992904588,
+      "loss": 3.4711,
+      "step": 53400
+    },
+    {
+      "epoch": 0.535,
+      "grad_norm": 0.08915389329195023,
+      "learning_rate": 0.0001933472978091327,
+      "loss": 3.4729,
+      "step": 53500
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 0.1250082552433014,
+      "learning_rate": 0.00019316685232156395,
+      "loss": 3.4727,
+      "step": 53600
+    },
+    {
+      "epoch": 0.537,
+      "grad_norm": 0.10531096905469894,
+      "learning_rate": 0.00019298691110661623,
+      "loss": 3.4734,
+      "step": 53700
+    },
+    {
+      "epoch": 0.538,
+      "grad_norm": 0.10072151571512222,
+      "learning_rate": 0.00019280747181992476,
+      "loss": 3.472,
+      "step": 53800
+    },
+    {
+      "epoch": 0.539,
+      "grad_norm": 0.08304274082183838,
+      "learning_rate": 0.0001926285321323549,
+      "loss": 3.4698,
+      "step": 53900
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.08655515313148499,
+      "learning_rate": 0.00019245008972987527,
+      "loss": 3.4707,
+      "step": 54000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 3.904296875,
+      "eval_runtime": 32.9698,
+      "eval_samples_per_second": 18264.853,
+      "eval_steps_per_second": 17.865,
+      "step": 54000
+    },
+    {
+      "epoch": 0.541,
+      "grad_norm": 0.09470842778682709,
+      "learning_rate": 0.00019227214231343207,
+      "loss": 3.4683,
+      "step": 54100
+    },
+    {
+      "epoch": 0.542,
+      "grad_norm": 0.09508795291185379,
+      "learning_rate": 0.00019209468759882463,
+      "loss": 3.4687,
+      "step": 54200
+    },
+    {
+      "epoch": 0.543,
+      "grad_norm": 0.09284261614084244,
+      "learning_rate": 0.00019191772331658236,
+      "loss": 3.4708,
+      "step": 54300
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.09701448678970337,
+      "learning_rate": 0.0001917412472118426,
+      "loss": 3.4713,
+      "step": 54400
+    },
+    {
+      "epoch": 0.545,
+      "grad_norm": 0.11029274761676788,
+      "learning_rate": 0.0001915652570442303,
+      "loss": 3.4722,
+      "step": 54500
+    },
+    {
+      "epoch": 0.546,
+      "grad_norm": 0.08805020898580551,
+      "learning_rate": 0.00019138975058773818,
+      "loss": 3.4685,
+      "step": 54600
+    },
+    {
+      "epoch": 0.547,
+      "grad_norm": 0.09119655936956406,
+      "learning_rate": 0.00019121472563060887,
+      "loss": 3.4686,
+      "step": 54700
+    },
+    {
+      "epoch": 0.548,
+      "grad_norm": 0.09098175168037415,
+      "learning_rate": 0.00019104017997521752,
+      "loss": 3.4695,
+      "step": 54800
+    },
+    {
+      "epoch": 0.549,
+      "grad_norm": 0.10574721544981003,
+      "learning_rate": 0.00019086611143795607,
+      "loss": 3.4681,
+      "step": 54900
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.09101998060941696,
+      "learning_rate": 0.00019069251784911847,
+      "loss": 3.4679,
+      "step": 55000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 3.90625,
+      "eval_runtime": 32.9737,
+      "eval_samples_per_second": 18262.678,
+      "eval_steps_per_second": 17.863,
+      "step": 55000
+    },
+    {
+      "epoch": 0.551,
+      "grad_norm": 0.09237556904554367,
+      "learning_rate": 0.00019051939705278708,
+      "loss": 3.4695,
+      "step": 55100
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 0.0976092666387558,
+      "learning_rate": 0.00019034674690672024,
+      "loss": 3.4685,
+      "step": 55200
+    },
+    {
+      "epoch": 0.553,
+      "grad_norm": 0.09669974446296692,
+      "learning_rate": 0.00019017456528224098,
+      "loss": 3.4668,
+      "step": 55300
+    },
+    {
+      "epoch": 0.554,
+      "grad_norm": 0.08793047815561295,
+      "learning_rate": 0.0001900028500641266,
+      "loss": 3.4681,
+      "step": 55400
+    },
+    {
+      "epoch": 0.555,
+      "grad_norm": 0.12426000833511353,
+      "learning_rate": 0.0001898315991504998,
+      "loss": 3.4674,
+      "step": 55500
+    },
+    {
+      "epoch": 0.556,
+      "grad_norm": 0.10039014369249344,
+      "learning_rate": 0.0001896608104527204,
+      "loss": 3.4688,
+      "step": 55600
+    },
+    {
+      "epoch": 0.557,
+      "grad_norm": 0.10602252185344696,
+      "learning_rate": 0.00018949048189527846,
+      "loss": 3.4674,
+      "step": 55700
+    },
+    {
+      "epoch": 0.558,
+      "grad_norm": 0.13479745388031006,
+      "learning_rate": 0.00018932061141568827,
+      "loss": 3.4679,
+      "step": 55800
+    },
+    {
+      "epoch": 0.559,
+      "grad_norm": 0.08916186541318893,
+      "learning_rate": 0.0001891511969643836,
+      "loss": 3.467,
+      "step": 55900
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12169528752565384,
+      "learning_rate": 0.0001889822365046136,
+      "loss": 3.4648,
+      "step": 56000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 3.90234375,
+      "eval_runtime": 33.0103,
+      "eval_samples_per_second": 18242.4,
+      "eval_steps_per_second": 17.843,
+      "step": 56000
+    },
+    {
+      "epoch": 0.561,
+      "grad_norm": 0.0936431810259819,
+      "learning_rate": 0.00018881372801234024,
+      "loss": 3.4669,
+      "step": 56100
+    },
+    {
+      "epoch": 0.562,
+      "grad_norm": 0.100394107401371,
+      "learning_rate": 0.00018864566947613624,
+      "loss": 3.4661,
+      "step": 56200
+    },
+    {
+      "epoch": 0.563,
+      "grad_norm": 0.0859285295009613,
+      "learning_rate": 0.00018847805889708435,
+      "loss": 3.4661,
+      "step": 56300
+    },
+    {
+      "epoch": 0.564,
+      "grad_norm": 0.08721830695867538,
+      "learning_rate": 0.00018831089428867736,
+      "loss": 3.4658,
+      "step": 56400
+    },
+    {
+      "epoch": 0.565,
+      "grad_norm": 0.15624579787254333,
+      "learning_rate": 0.00018814417367671947,
+      "loss": 3.4653,
+      "step": 56500
+    },
+    {
+      "epoch": 0.566,
+      "grad_norm": 0.09214714914560318,
+      "learning_rate": 0.00018797789509922812,
+      "loss": 3.4642,
+      "step": 56600
+    },
+    {
+      "epoch": 0.567,
+      "grad_norm": 0.13581374287605286,
+      "learning_rate": 0.000187812056606337,
+      "loss": 3.4637,
+      "step": 56700
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 0.09718205779790878,
+      "learning_rate": 0.0001876466562602004,
+      "loss": 3.4657,
+      "step": 56800
+    },
+    {
+      "epoch": 0.569,
+      "grad_norm": 0.09438052028417587,
+      "learning_rate": 0.00018748169213489755,
+      "loss": 3.4671,
+      "step": 56900
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.090947225689888,
+      "learning_rate": 0.0001873171623163388,
+      "loss": 3.4655,
+      "step": 57000
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 3.900390625,
+      "eval_runtime": 32.9917,
+      "eval_samples_per_second": 18252.689,
+      "eval_steps_per_second": 17.853,
+      "step": 57000
+    },
+    {
+      "epoch": 0.571,
+      "grad_norm": 0.09600440412759781,
+      "learning_rate": 0.0001871530649021723,
+      "loss": 3.4676,
+      "step": 57100
+    },
+    {
+      "epoch": 0.572,
+      "grad_norm": 0.09546720236539841,
+      "learning_rate": 0.00018698939800169143,
+      "loss": 3.4651,
+      "step": 57200
+    },
+    {
+      "epoch": 0.573,
+      "grad_norm": 0.11584016680717468,
+      "learning_rate": 0.0001868261597357436,
+      "loss": 3.4654,
+      "step": 57300
+    },
+    {
+      "epoch": 0.574,
+      "grad_norm": 0.09096308052539825,
+      "learning_rate": 0.00018666334823663937,
+      "loss": 3.4642,
+      "step": 57400
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 0.09048023074865341,
+      "learning_rate": 0.00018650096164806276,
+      "loss": 3.4632,
+      "step": 57500
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.08897216618061066,
+      "learning_rate": 0.0001863389981249825,
+      "loss": 3.4642,
+      "step": 57600
+    },
+    {
+      "epoch": 0.577,
+      "grad_norm": 0.08574885874986649,
+      "learning_rate": 0.00018617745583356373,
+      "loss": 3.4645,
+      "step": 57700
+    },
+    {
+      "epoch": 0.578,
+      "grad_norm": 0.09016621857881546,
+      "learning_rate": 0.00018601633295108115,
+      "loss": 3.4651,
+      "step": 57800
+    },
+    {
+      "epoch": 0.579,
+      "grad_norm": 0.08941857516765594,
+      "learning_rate": 0.0001858556276658322,
+      "loss": 3.4656,
+      "step": 57900
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.08731842786073685,
+      "learning_rate": 0.00018569533817705186,
+      "loss": 3.4664,
+      "step": 58000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 3.896484375,
+      "eval_runtime": 32.9962,
+      "eval_samples_per_second": 18250.233,
+      "eval_steps_per_second": 17.851,
+      "step": 58000
+    },
+    {
+      "epoch": 0.581,
+      "grad_norm": 0.09401644766330719,
+      "learning_rate": 0.00018553546269482776,
+      "loss": 3.462,
+      "step": 58100
+    },
+    {
+      "epoch": 0.582,
+      "grad_norm": 0.1261298507452011,
+      "learning_rate": 0.00018537599944001617,
+      "loss": 3.4631,
+      "step": 58200
+    },
+    {
+      "epoch": 0.583,
+      "grad_norm": 0.08777070790529251,
+      "learning_rate": 0.00018521694664415904,
+      "loss": 3.4647,
+      "step": 58300
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 0.10590939223766327,
+      "learning_rate": 0.00018505830254940132,
+      "loss": 3.4645,
+      "step": 58400
+    },
+    {
+      "epoch": 0.585,
+      "grad_norm": 0.08615025877952576,
+      "learning_rate": 0.0001849000654084097,
+      "loss": 3.463,
+      "step": 58500
+    },
+    {
+      "epoch": 0.586,
+      "grad_norm": 0.09571292251348495,
+      "learning_rate": 0.00018474223348429158,
+      "loss": 3.4624,
+      "step": 58600
+    },
+    {
+      "epoch": 0.587,
+      "grad_norm": 0.08672121912240982,
+      "learning_rate": 0.000184584805050515,
+      "loss": 3.4636,
+      "step": 58700
+    },
+    {
+      "epoch": 0.588,
+      "grad_norm": 0.09722130000591278,
+      "learning_rate": 0.00018442777839082936,
+      "loss": 3.4615,
+      "step": 58800
+    },
+    {
+      "epoch": 0.589,
+      "grad_norm": 0.08235087245702744,
+      "learning_rate": 0.00018427115179918692,
+      "loss": 3.4631,
+      "step": 58900
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.11260483413934708,
+      "learning_rate": 0.00018411492357966468,
+      "loss": 3.46,
+      "step": 59000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 3.8984375,
+      "eval_runtime": 32.986,
+      "eval_samples_per_second": 18255.851,
+      "eval_steps_per_second": 17.856,
+      "step": 59000
+    },
+    {
+      "epoch": 0.591,
+      "grad_norm": 0.09491027891635895,
+      "learning_rate": 0.00018395909204638758,
+      "loss": 3.4612,
+      "step": 59100
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.09216572344303131,
+      "learning_rate": 0.00018380365552345197,
+      "loss": 3.4632,
+      "step": 59200
+    },
+    {
+      "epoch": 0.593,
+      "grad_norm": 0.09314817935228348,
+      "learning_rate": 0.00018364861234484967,
+      "loss": 3.4616,
+      "step": 59300
+    },
+    {
+      "epoch": 0.594,
+      "grad_norm": 0.10169938951730728,
+      "learning_rate": 0.00018349396085439343,
+      "loss": 3.4638,
+      "step": 59400
+    },
+    {
+      "epoch": 0.595,
+      "grad_norm": 0.10830460488796234,
+      "learning_rate": 0.00018333969940564226,
+      "loss": 3.4622,
+      "step": 59500
+    },
+    {
+      "epoch": 0.596,
+      "grad_norm": 0.09347419440746307,
+      "learning_rate": 0.00018318582636182794,
+      "loss": 3.461,
+      "step": 59600
+    },
+    {
+      "epoch": 0.597,
+      "grad_norm": 0.09264504164457321,
+      "learning_rate": 0.00018303234009578204,
+      "loss": 3.4604,
+      "step": 59700
+    },
+    {
+      "epoch": 0.598,
+      "grad_norm": 0.08877206593751907,
+      "learning_rate": 0.00018287923898986378,
+      "loss": 3.4622,
+      "step": 59800
+    },
+    {
+      "epoch": 0.599,
+      "grad_norm": 0.09566622227430344,
+      "learning_rate": 0.00018272652143588817,
+      "loss": 3.4635,
+      "step": 59900
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10193250328302383,
+      "learning_rate": 0.00018257418583505537,
+      "loss": 3.4608,
+      "step": 60000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 3.8984375,
+      "eval_runtime": 32.9947,
+      "eval_samples_per_second": 18251.046,
+      "eval_steps_per_second": 17.851,
+      "step": 60000
+    },
+    {
+      "epoch": 0.601,
+      "grad_norm": 0.09149635583162308,
+      "learning_rate": 0.00018242223059788013,
+      "loss": 3.4603,
+      "step": 60100
+    },
+    {
+      "epoch": 0.602,
+      "grad_norm": 0.09358430653810501,
+      "learning_rate": 0.0001822706541441223,
+      "loss": 3.4595,
+      "step": 60200
+    },
+    {
+      "epoch": 0.603,
+      "grad_norm": 0.10326792299747467,
+      "learning_rate": 0.00018211945490271766,
+      "loss": 3.4619,
+      "step": 60300
+    },
+    {
+      "epoch": 0.604,
+      "grad_norm": 0.10516923666000366,
+      "learning_rate": 0.00018196863131170974,
+      "loss": 3.4592,
+      "step": 60400
+    },
+    {
+      "epoch": 0.605,
+      "grad_norm": 0.09201698005199432,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 3.4594,
+      "step": 60500
+    },
+    {
+      "epoch": 0.606,
+      "grad_norm": 0.12762495875358582,
+      "learning_rate": 0.00018166810487818988,
+      "loss": 3.4608,
+      "step": 60600
+    },
+    {
+      "epoch": 0.607,
+      "grad_norm": 0.09665093570947647,
+      "learning_rate": 0.00018151839895669614,
+      "loss": 3.4579,
+      "step": 60700
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.0870705246925354,
+      "learning_rate": 0.00018136906252750294,
+      "loss": 3.4588,
+      "step": 60800
+    },
+    {
+      "epoch": 0.609,
+      "grad_norm": 0.0998244658112526,
+      "learning_rate": 0.00018122009407318745,
+      "loss": 3.4589,
+      "step": 60900
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.08676362037658691,
+      "learning_rate": 0.00018107149208503707,
+      "loss": 3.4604,
+      "step": 61000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 3.892578125,
+      "eval_runtime": 33.0945,
+      "eval_samples_per_second": 18195.999,
+      "eval_steps_per_second": 17.798,
+      "step": 61000
+    },
+    {
+      "epoch": 0.611,
+      "grad_norm": 0.09438201040029526,
+      "learning_rate": 0.000180923255062985,
+      "loss": 3.4588,
+      "step": 61100
+    },
+    {
+      "epoch": 0.612,
+      "grad_norm": 0.09467477351427078,
+      "learning_rate": 0.0001807753815155468,
+      "loss": 3.4595,
+      "step": 61200
+    },
+    {
+      "epoch": 0.613,
+      "grad_norm": 0.09366316348314285,
+      "learning_rate": 0.00018062786995975738,
+      "loss": 3.4596,
+      "step": 61300
+    },
+    {
+      "epoch": 0.614,
+      "grad_norm": 0.09797662496566772,
+      "learning_rate": 0.0001804807189211084,
+      "loss": 3.46,
+      "step": 61400
+    },
+    {
+      "epoch": 0.615,
+      "grad_norm": 0.09239408373832703,
+      "learning_rate": 0.00018033392693348645,
+      "loss": 3.4592,
+      "step": 61500
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 0.1170559823513031,
+      "learning_rate": 0.0001801874925391118,
+      "loss": 3.4594,
+      "step": 61600
+    },
+    {
+      "epoch": 0.617,
+      "grad_norm": 0.11294198781251907,
+      "learning_rate": 0.00018004141428847736,
+      "loss": 3.4566,
+      "step": 61700
+    },
+    {
+      "epoch": 0.618,
+      "grad_norm": 0.08425901085138321,
+      "learning_rate": 0.00017989569074028863,
+      "loss": 3.4582,
+      "step": 61800
+    },
+    {
+      "epoch": 0.619,
+      "grad_norm": 0.08901657909154892,
+      "learning_rate": 0.0001797503204614039,
+      "loss": 3.4575,
+      "step": 61900
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.09452513605356216,
+      "learning_rate": 0.00017960530202677493,
+      "loss": 3.4584,
+      "step": 62000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 3.888671875,
+      "eval_runtime": 33.0119,
+      "eval_samples_per_second": 18241.555,
+      "eval_steps_per_second": 17.842,
+      "step": 62000
+    },
+    {
+      "epoch": 0.621,
+      "grad_norm": 0.10306137800216675,
+      "learning_rate": 0.0001794606340193885,
+      "loss": 3.4585,
+      "step": 62100
+    },
+    {
+      "epoch": 0.622,
+      "grad_norm": 0.09906180948019028,
+      "learning_rate": 0.00017931631503020814,
+      "loss": 3.4559,
+      "step": 62200
+    },
+    {
+      "epoch": 0.623,
+      "grad_norm": 0.09659173339605331,
+      "learning_rate": 0.00017917234365811654,
+      "loss": 3.4568,
+      "step": 62300
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.08893030881881714,
+      "learning_rate": 0.0001790287185098582,
+      "loss": 3.4596,
+      "step": 62400
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.09648360311985016,
+      "learning_rate": 0.00017888543819998318,
+      "loss": 3.4573,
+      "step": 62500
+    },
+    {
+      "epoch": 0.626,
+      "grad_norm": 0.08842136710882187,
+      "learning_rate": 0.0001787425013507906,
+      "loss": 3.4575,
+      "step": 62600
+    },
+    {
+      "epoch": 0.627,
+      "grad_norm": 0.097404845058918,
+      "learning_rate": 0.00017859990659227327,
+      "loss": 3.4553,
+      "step": 62700
+    },
+    {
+      "epoch": 0.628,
+      "grad_norm": 0.10879901051521301,
+      "learning_rate": 0.0001784576525620624,
+      "loss": 3.4557,
+      "step": 62800
+    },
+    {
+      "epoch": 0.629,
+      "grad_norm": 0.08811251074075699,
+      "learning_rate": 0.000178315737905373,
+      "loss": 3.4547,
+      "step": 62900
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.09087971597909927,
+      "learning_rate": 0.0001781741612749496,
+      "loss": 3.4556,
+      "step": 63000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 3.89453125,
+      "eval_runtime": 33.0103,
+      "eval_samples_per_second": 18242.415,
+      "eval_steps_per_second": 17.843,
+      "step": 63000
+    },
+    {
+      "epoch": 0.631,
+      "grad_norm": 0.08818445354700089,
+      "learning_rate": 0.0001780329213310126,
+      "loss": 3.4553,
+      "step": 63100
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 0.09016218781471252,
+      "learning_rate": 0.000177892016741205,
+      "loss": 3.4546,
+      "step": 63200
+    },
+    {
+      "epoch": 0.633,
+      "grad_norm": 0.08760054409503937,
+      "learning_rate": 0.0001777514461805397,
+      "loss": 3.4548,
+      "step": 63300
+    },
+    {
+      "epoch": 0.634,
+      "grad_norm": 0.08488897979259491,
+      "learning_rate": 0.00017761120833134698,
+      "loss": 3.4575,
+      "step": 63400
+    },
+    {
+      "epoch": 0.635,
+      "grad_norm": 0.09008808434009552,
+      "learning_rate": 0.00017747130188322277,
+      "loss": 3.4554,
+      "step": 63500
+    },
+    {
+      "epoch": 0.636,
+      "grad_norm": 0.09507942944765091,
+      "learning_rate": 0.00017733172553297715,
+      "loss": 3.4548,
+      "step": 63600
+    },
+    {
+      "epoch": 0.637,
+      "grad_norm": 0.08736253529787064,
+      "learning_rate": 0.00017719247798458348,
+      "loss": 3.4547,
+      "step": 63700
+    },
+    {
+      "epoch": 0.638,
+      "grad_norm": 0.1137012243270874,
+      "learning_rate": 0.00017705355794912778,
+      "loss": 3.4586,
+      "step": 63800
+    },
+    {
+      "epoch": 0.639,
+      "grad_norm": 0.100893035531044,
+      "learning_rate": 0.00017691496414475844,
+      "loss": 3.4563,
+      "step": 63900
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.0883665531873703,
+      "learning_rate": 0.00017677669529663688,
+      "loss": 3.4545,
+      "step": 64000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 3.90234375,
+      "eval_runtime": 32.9836,
+      "eval_samples_per_second": 18257.192,
+      "eval_steps_per_second": 17.857,
+      "step": 64000
+    },
+    {
+      "epoch": 0.641,
+      "grad_norm": 0.09608853608369827,
+      "learning_rate": 0.00017663875013688816,
+      "loss": 3.4548,
+      "step": 64100
+    },
+    {
+      "epoch": 0.642,
+      "grad_norm": 0.08302801102399826,
+      "learning_rate": 0.00017650112740455194,
+      "loss": 3.454,
+      "step": 64200
+    },
+    {
+      "epoch": 0.643,
+      "grad_norm": 0.08717923611402512,
+      "learning_rate": 0.0001763638258455345,
+      "loss": 3.4555,
+      "step": 64300
+    },
+    {
+      "epoch": 0.644,
+      "grad_norm": 0.11070983111858368,
+      "learning_rate": 0.00017622684421256035,
+      "loss": 3.4551,
+      "step": 64400
+    },
+    {
+      "epoch": 0.645,
+      "grad_norm": 0.10088195651769638,
+      "learning_rate": 0.00017609018126512477,
+      "loss": 3.4545,
+      "step": 64500
+    },
+    {
+      "epoch": 0.646,
+      "grad_norm": 0.08371099084615707,
+      "learning_rate": 0.00017595383576944672,
+      "loss": 3.4552,
+      "step": 64600
+    },
+    {
+      "epoch": 0.647,
+      "grad_norm": 0.08482113480567932,
+      "learning_rate": 0.00017581780649842193,
+      "loss": 3.4547,
+      "step": 64700
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 0.10641901195049286,
+      "learning_rate": 0.00017568209223157663,
+      "loss": 3.4539,
+      "step": 64800
+    },
+    {
+      "epoch": 0.649,
+      "grad_norm": 0.12532317638397217,
+      "learning_rate": 0.00017554669175502143,
+      "loss": 3.4542,
+      "step": 64900
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.10789231210947037,
+      "learning_rate": 0.00017541160386140586,
+      "loss": 3.4526,
+      "step": 65000
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 3.884765625,
+      "eval_runtime": 32.991,
+      "eval_samples_per_second": 18253.083,
+      "eval_steps_per_second": 17.853,
+      "step": 65000
+    },
+    {
+      "epoch": 0.651,
+      "grad_norm": 0.10503752529621124,
+      "learning_rate": 0.00017527682734987297,
+      "loss": 3.4551,
+      "step": 65100
+    },
+    {
+      "epoch": 0.652,
+      "grad_norm": 0.08596997708082199,
+      "learning_rate": 0.00017514236102601468,
+      "loss": 3.4544,
+      "step": 65200
+    },
+    {
+      "epoch": 0.653,
+      "grad_norm": 0.10656549036502838,
+      "learning_rate": 0.00017500820370182729,
+      "loss": 3.452,
+      "step": 65300
+    },
+    {
+      "epoch": 0.654,
+      "grad_norm": 0.10545896738767624,
+      "learning_rate": 0.00017487435419566726,
+      "loss": 3.4525,
+      "step": 65400
+    },
+    {
+      "epoch": 0.655,
+      "grad_norm": 0.10238152742385864,
+      "learning_rate": 0.0001747408113322076,
+      "loss": 3.453,
+      "step": 65500
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.1159452497959137,
+      "learning_rate": 0.00017460757394239458,
+      "loss": 3.4533,
+      "step": 65600
+    },
+    {
+      "epoch": 0.657,
+      "grad_norm": 0.08971402049064636,
+      "learning_rate": 0.00017447464086340456,
+      "loss": 3.4525,
+      "step": 65700
+    },
+    {
+      "epoch": 0.658,
+      "grad_norm": 0.09728628396987915,
+      "learning_rate": 0.00017434201093860166,
+      "loss": 3.4533,
+      "step": 65800
+    },
+    {
+      "epoch": 0.659,
+      "grad_norm": 0.10055437684059143,
+      "learning_rate": 0.00017420968301749517,
+      "loss": 3.4515,
+      "step": 65900
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.10009339451789856,
+      "learning_rate": 0.00017407765595569785,
+      "loss": 3.4532,
+      "step": 66000
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 3.888671875,
+      "eval_runtime": 33.007,
+      "eval_samples_per_second": 18244.268,
+      "eval_steps_per_second": 17.845,
+      "step": 66000
+    },
+    {
+      "epoch": 0.661,
+      "grad_norm": 0.08580023795366287,
+      "learning_rate": 0.00017394592861488424,
+      "loss": 3.4505,
+      "step": 66100
+    },
+    {
+      "epoch": 0.662,
+      "grad_norm": 0.08647996932268143,
+      "learning_rate": 0.00017381449986274955,
+      "loss": 3.4532,
+      "step": 66200
+    },
+    {
+      "epoch": 0.663,
+      "grad_norm": 0.08780326694250107,
+      "learning_rate": 0.00017368336857296875,
+      "loss": 3.4515,
+      "step": 66300
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 0.09744898974895477,
+      "learning_rate": 0.00017355253362515583,
+      "loss": 3.4513,
+      "step": 66400
+    },
+    {
+      "epoch": 0.665,
+      "grad_norm": 0.10528256744146347,
+      "learning_rate": 0.000173421993904824,
+      "loss": 3.4534,
+      "step": 66500
+    },
+    {
+      "epoch": 0.666,
+      "grad_norm": 0.0887388065457344,
+      "learning_rate": 0.00017329174830334545,
+      "loss": 3.4532,
+      "step": 66600
+    },
+    {
+      "epoch": 0.667,
+      "grad_norm": 0.11578360199928284,
+      "learning_rate": 0.00017316179571791196,
+      "loss": 3.4502,
+      "step": 66700
+    },
+    {
+      "epoch": 0.668,
+      "grad_norm": 0.09660936146974564,
+      "learning_rate": 0.0001730321350514957,
+      "loss": 3.4531,
+      "step": 66800
+    },
+    {
+      "epoch": 0.669,
+      "grad_norm": 0.09446755796670914,
+      "learning_rate": 0.0001729027652128102,
+      "loss": 3.4543,
+      "step": 66900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.10190019011497498,
+      "learning_rate": 0.00017277368511627204,
+      "loss": 3.4509,
+      "step": 67000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 3.8828125,
+      "eval_runtime": 32.992,
+      "eval_samples_per_second": 18252.544,
+      "eval_steps_per_second": 17.853,
+      "step": 67000
+    },
+    {
+      "epoch": 0.671,
+      "grad_norm": 0.10701554268598557,
+      "learning_rate": 0.0001726448936819622,
+      "loss": 3.4513,
+      "step": 67100
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 0.11148486286401749,
+      "learning_rate": 0.00017251638983558853,
+      "loss": 3.4505,
+      "step": 67200
+    },
+    {
+      "epoch": 0.673,
+      "grad_norm": 0.10375931113958359,
+      "learning_rate": 0.00017238817250844786,
+      "loss": 3.4514,
+      "step": 67300
+    },
+    {
+      "epoch": 0.674,
+      "grad_norm": 0.09482084214687347,
+      "learning_rate": 0.00017226024063738863,
+      "loss": 3.4498,
+      "step": 67400
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 0.11015119403600693,
+      "learning_rate": 0.00017213259316477408,
+      "loss": 3.4503,
+      "step": 67500
+    },
+    {
+      "epoch": 0.676,
+      "grad_norm": 0.09139025211334229,
+      "learning_rate": 0.00017200522903844536,
+      "loss": 3.4503,
+      "step": 67600
+    },
+    {
+      "epoch": 0.677,
+      "grad_norm": 0.08138120919466019,
+      "learning_rate": 0.00017187814721168515,
+      "loss": 3.4496,
+      "step": 67700
+    },
+    {
+      "epoch": 0.678,
+      "grad_norm": 0.11369217187166214,
+      "learning_rate": 0.00017175134664318157,
+      "loss": 3.4493,
+      "step": 67800
+    },
+    {
+      "epoch": 0.679,
+      "grad_norm": 0.09752605855464935,
+      "learning_rate": 0.00017162482629699222,
+      "loss": 3.4484,
+      "step": 67900
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.0917077511548996,
+      "learning_rate": 0.00017149858514250882,
+      "loss": 3.4493,
+      "step": 68000
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 3.880859375,
+      "eval_runtime": 32.9961,
+      "eval_samples_per_second": 18250.289,
+      "eval_steps_per_second": 17.851,
+      "step": 68000
+    },
+    {
+      "epoch": 0.681,
+      "grad_norm": 0.09174978733062744,
+      "learning_rate": 0.00017137262215442185,
+      "loss": 3.45,
+      "step": 68100
+    },
+    {
+      "epoch": 0.682,
+      "grad_norm": 0.11849396675825119,
+      "learning_rate": 0.00017124693631268543,
+      "loss": 3.4511,
+      "step": 68200
+    },
+    {
+      "epoch": 0.683,
+      "grad_norm": 0.0938100814819336,
+      "learning_rate": 0.00017112152660248296,
+      "loss": 3.4496,
+      "step": 68300
+    },
+    {
+      "epoch": 0.684,
+      "grad_norm": 0.11769715696573257,
+      "learning_rate": 0.00017099639201419238,
+      "loss": 3.4495,
+      "step": 68400
+    },
+    {
+      "epoch": 0.685,
+      "grad_norm": 0.11145388334989548,
+      "learning_rate": 0.0001708715315433522,
+      "loss": 3.4493,
+      "step": 68500
+    },
+    {
+      "epoch": 0.686,
+      "grad_norm": 0.12704512476921082,
+      "learning_rate": 0.00017074694419062767,
+      "loss": 3.4503,
+      "step": 68600
+    },
+    {
+      "epoch": 0.687,
+      "grad_norm": 0.10009263455867767,
+      "learning_rate": 0.00017062262896177705,
+      "loss": 3.4527,
+      "step": 68700
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 0.0949656069278717,
+      "learning_rate": 0.00017049858486761839,
+      "loss": 3.4497,
+      "step": 68800
+    },
+    {
+      "epoch": 0.689,
+      "grad_norm": 0.08776091784238815,
+      "learning_rate": 0.0001703748109239964,
+      "loss": 3.4493,
+      "step": 68900
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12844863533973694,
+      "learning_rate": 0.0001702513061517497,
+      "loss": 3.449,
+      "step": 69000
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 3.8828125,
+      "eval_runtime": 33.0055,
+      "eval_samples_per_second": 18245.099,
+      "eval_steps_per_second": 17.846,
+      "step": 69000
+    },
+    {
+      "epoch": 0.691,
+      "grad_norm": 0.086809441447258,
+      "learning_rate": 0.0001701280695766784,
+      "loss": 3.4488,
+      "step": 69100
+    },
+    {
+      "epoch": 0.692,
+      "grad_norm": 0.09244389086961746,
+      "learning_rate": 0.0001700051002295115,
+      "loss": 3.4473,
+      "step": 69200
+    },
+    {
+      "epoch": 0.693,
+      "grad_norm": 0.08996045589447021,
+      "learning_rate": 0.00016988239714587518,
+      "loss": 3.4499,
+      "step": 69300
+    },
+    {
+      "epoch": 0.694,
+      "grad_norm": 0.10364508628845215,
+      "learning_rate": 0.00016975995936626098,
+      "loss": 3.4495,
+      "step": 69400
+    },
+    {
+      "epoch": 0.695,
+      "grad_norm": 0.09321583062410355,
+      "learning_rate": 0.0001696377859359942,
+      "loss": 3.4468,
+      "step": 69500
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 0.09319949150085449,
+      "learning_rate": 0.0001695158759052026,
+      "loss": 3.447,
+      "step": 69600
+    },
+    {
+      "epoch": 0.697,
+      "grad_norm": 0.0915616974234581,
+      "learning_rate": 0.00016939422832878555,
+      "loss": 3.4461,
+      "step": 69700
+    },
+    {
+      "epoch": 0.698,
+      "grad_norm": 0.08925987035036087,
+      "learning_rate": 0.00016927284226638315,
+      "loss": 3.4479,
+      "step": 69800
+    },
+    {
+      "epoch": 0.699,
+      "grad_norm": 0.10480326414108276,
+      "learning_rate": 0.00016915171678234564,
+      "loss": 3.4477,
+      "step": 69900
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.09662549942731857,
+      "learning_rate": 0.0001690308509457033,
+      "loss": 3.4464,
+      "step": 70000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 3.87890625,
+      "eval_runtime": 33.1572,
+      "eval_samples_per_second": 18161.615,
+      "eval_steps_per_second": 17.764,
+      "step": 70000
+    },
+    {
+      "epoch": 0.701,
+      "grad_norm": 0.08479616045951843,
+      "learning_rate": 0.0001689102438301362,
+      "loss": 3.4458,
+      "step": 70100
+    },
+    {
+      "epoch": 0.702,
+      "grad_norm": 0.09455426782369614,
+      "learning_rate": 0.00016878989451394444,
+      "loss": 3.447,
+      "step": 70200
+    },
+    {
+      "epoch": 0.703,
+      "grad_norm": 0.08542288839817047,
+      "learning_rate": 0.00016866980208001865,
+      "loss": 3.4462,
+      "step": 70300
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 0.10905344784259796,
+      "learning_rate": 0.00016854996561581053,
+      "loss": 3.4454,
+      "step": 70400
+    },
+    {
+      "epoch": 0.705,
+      "grad_norm": 0.0927351713180542,
+      "learning_rate": 0.00016843038421330382,
+      "loss": 3.4444,
+      "step": 70500
+    },
+    {
+      "epoch": 0.706,
+      "grad_norm": 0.0991898775100708,
+      "learning_rate": 0.00016831105696898528,
+      "loss": 3.4476,
+      "step": 70600
+    },
+    {
+      "epoch": 0.707,
+      "grad_norm": 0.07934626936912537,
+      "learning_rate": 0.0001681919829838161,
+      "loss": 3.4465,
+      "step": 70700
+    },
+    {
+      "epoch": 0.708,
+      "grad_norm": 0.0833369791507721,
+      "learning_rate": 0.0001680731613632036,
+      "loss": 3.4482,
+      "step": 70800
+    },
+    {
+      "epoch": 0.709,
+      "grad_norm": 0.09017114341259003,
+      "learning_rate": 0.00016795459121697257,
+      "loss": 3.4453,
+      "step": 70900
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.09019383788108826,
+      "learning_rate": 0.00016783627165933783,
+      "loss": 3.446,
+      "step": 71000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 3.890625,
+      "eval_runtime": 33.0104,
+      "eval_samples_per_second": 18242.397,
+      "eval_steps_per_second": 17.843,
+      "step": 71000
+    },
+    {
+      "epoch": 0.711,
+      "grad_norm": 0.09486021101474762,
+      "learning_rate": 0.0001677182018088759,
+      "loss": 3.4462,
+      "step": 71100
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 0.09535174071788788,
+      "learning_rate": 0.00016760038078849775,
+      "loss": 3.4483,
+      "step": 71200
+    },
+    {
+      "epoch": 0.713,
+      "grad_norm": 0.08906789124011993,
+      "learning_rate": 0.00016748280772542137,
+      "loss": 3.445,
+      "step": 71300
+    },
+    {
+      "epoch": 0.714,
+      "grad_norm": 0.09585762023925781,
+      "learning_rate": 0.0001673654817511446,
+      "loss": 3.4469,
+      "step": 71400
+    },
+    {
+      "epoch": 0.715,
+      "grad_norm": 0.10558847337961197,
+      "learning_rate": 0.00016724840200141815,
+      "loss": 3.4462,
+      "step": 71500
+    },
+    {
+      "epoch": 0.716,
+      "grad_norm": 0.09401770681142807,
+      "learning_rate": 0.0001671315676162189,
+      "loss": 3.4436,
+      "step": 71600
+    },
+    {
+      "epoch": 0.717,
+      "grad_norm": 0.09375085681676865,
+      "learning_rate": 0.00016701497773972333,
+      "loss": 3.445,
+      "step": 71700
+    },
+    {
+      "epoch": 0.718,
+      "grad_norm": 0.09521766006946564,
+      "learning_rate": 0.00016689863152028126,
+      "loss": 3.443,
+      "step": 71800
+    },
+    {
+      "epoch": 0.719,
+      "grad_norm": 0.09246081113815308,
+      "learning_rate": 0.00016678252811038962,
+      "loss": 3.4458,
+      "step": 71900
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.08197703957557678,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 3.4456,
+      "step": 72000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 3.880859375,
+      "eval_runtime": 32.9885,
+      "eval_samples_per_second": 18254.48,
+      "eval_steps_per_second": 17.855,
+      "step": 72000
+    },
+    {
+      "epoch": 0.721,
+      "grad_norm": 0.08918161690235138,
+      "learning_rate": 0.00016655104634982608,
+      "loss": 3.4453,
+      "step": 72100
+    },
+    {
+      "epoch": 0.722,
+      "grad_norm": 0.08643663674592972,
+      "learning_rate": 0.00016643566632465153,
+      "loss": 3.4471,
+      "step": 72200
+    },
+    {
+      "epoch": 0.723,
+      "grad_norm": 0.08612331748008728,
+      "learning_rate": 0.0001663205257599714,
+      "loss": 3.4442,
+      "step": 72300
+    },
+    {
+      "epoch": 0.724,
+      "grad_norm": 0.11025555431842804,
+      "learning_rate": 0.00016620562382863342,
+      "loss": 3.4443,
+      "step": 72400
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 0.09381308406591415,
+      "learning_rate": 0.00016609095970747994,
+      "loss": 3.4414,
+      "step": 72500
+    },
+    {
+      "epoch": 0.726,
+      "grad_norm": 0.094405896961689,
+      "learning_rate": 0.00016597653257732306,
+      "loss": 3.4438,
+      "step": 72600
+    },
+    {
+      "epoch": 0.727,
+      "grad_norm": 0.09081803262233734,
+      "learning_rate": 0.00016586234162292005,
+      "loss": 3.444,
+      "step": 72700
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 0.11879288405179977,
+      "learning_rate": 0.00016574838603294897,
+      "loss": 3.4418,
+      "step": 72800
+    },
+    {
+      "epoch": 0.729,
+      "grad_norm": 0.09423980116844177,
+      "learning_rate": 0.00016563466499998442,
+      "loss": 3.4443,
+      "step": 72900
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.0863361731171608,
+      "learning_rate": 0.00016552117772047358,
+      "loss": 3.4436,
+      "step": 73000
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 3.884765625,
+      "eval_runtime": 33.0049,
+      "eval_samples_per_second": 18245.429,
+      "eval_steps_per_second": 17.846,
+      "step": 73000
+    },
+    {
+      "epoch": 0.731,
+      "grad_norm": 0.13333062827587128,
+      "learning_rate": 0.00016540792339471237,
+      "loss": 3.4447,
+      "step": 73100
+    },
+    {
+      "epoch": 0.732,
+      "grad_norm": 0.0957045704126358,
+      "learning_rate": 0.00016529490122682157,
+      "loss": 3.4425,
+      "step": 73200
+    },
+    {
+      "epoch": 0.733,
+      "grad_norm": 0.12515659630298615,
+      "learning_rate": 0.00016518211042472372,
+      "loss": 3.4423,
+      "step": 73300
+    },
+    {
+      "epoch": 0.734,
+      "grad_norm": 0.09493029862642288,
+      "learning_rate": 0.00016506955020011946,
+      "loss": 3.4401,
+      "step": 73400
+    },
+    {
+      "epoch": 0.735,
+      "grad_norm": 0.08881188184022903,
+      "learning_rate": 0.0001649572197684645,
+      "loss": 3.4419,
+      "step": 73500
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 0.10456575453281403,
+      "learning_rate": 0.00016484511834894677,
+      "loss": 3.4428,
+      "step": 73600
+    },
+    {
+      "epoch": 0.737,
+      "grad_norm": 0.08246786147356033,
+      "learning_rate": 0.00016473324516446338,
+      "loss": 3.4445,
+      "step": 73700
+    },
+    {
+      "epoch": 0.738,
+      "grad_norm": 0.11760751903057098,
+      "learning_rate": 0.00016462159944159827,
+      "loss": 3.4431,
+      "step": 73800
+    },
+    {
+      "epoch": 0.739,
+      "grad_norm": 0.09637313336133957,
+      "learning_rate": 0.00016451018041059955,
+      "loss": 3.4426,
+      "step": 73900
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.13058100640773773,
+      "learning_rate": 0.0001643989873053573,
+      "loss": 3.4425,
+      "step": 74000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 3.8828125,
+      "eval_runtime": 33.0139,
+      "eval_samples_per_second": 18240.416,
+      "eval_steps_per_second": 17.841,
+      "step": 74000
+    },
+    {
+      "epoch": 0.741,
+      "grad_norm": 0.08998531848192215,
+      "learning_rate": 0.00016428801936338143,
+      "loss": 3.4437,
+      "step": 74100
+    },
+    {
+      "epoch": 0.742,
+      "grad_norm": 0.09501896053552628,
+      "learning_rate": 0.00016417727582577964,
+      "loss": 3.4427,
+      "step": 74200
+    },
+    {
+      "epoch": 0.743,
+      "grad_norm": 0.10081542283296585,
+      "learning_rate": 0.00016406675593723583,
+      "loss": 3.4416,
+      "step": 74300
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 0.08788462728261948,
+      "learning_rate": 0.00016395645894598824,
+      "loss": 3.4433,
+      "step": 74400
+    },
+    {
+      "epoch": 0.745,
+      "grad_norm": 0.13512863218784332,
+      "learning_rate": 0.0001638463841038081,
+      "loss": 3.443,
+      "step": 74500
+    },
+    {
+      "epoch": 0.746,
+      "grad_norm": 0.08741659671068192,
+      "learning_rate": 0.00016373653066597825,
+      "loss": 3.4422,
+      "step": 74600
+    },
+    {
+      "epoch": 0.747,
+      "grad_norm": 0.09418010711669922,
+      "learning_rate": 0.000163626897891272,
+      "loss": 3.4409,
+      "step": 74700
+    },
+    {
+      "epoch": 0.748,
+      "grad_norm": 0.09971028566360474,
+      "learning_rate": 0.00016351748504193217,
+      "loss": 3.4415,
+      "step": 74800
+    },
+    {
+      "epoch": 0.749,
+      "grad_norm": 0.10280565172433853,
+      "learning_rate": 0.0001634082913836501,
+      "loss": 3.4405,
+      "step": 74900
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.09686677157878876,
+      "learning_rate": 0.00016329931618554522,
+      "loss": 3.4416,
+      "step": 75000
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 3.875,
+      "eval_runtime": 32.9976,
+      "eval_samples_per_second": 18249.452,
+      "eval_steps_per_second": 17.85,
+      "step": 75000
+    },
+    {
+      "epoch": 0.751,
+      "grad_norm": 0.08808457851409912,
+      "learning_rate": 0.00016319055872014417,
+      "loss": 3.4405,
+      "step": 75100
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 0.09505708515644073,
+      "learning_rate": 0.00016308201826336055,
+      "loss": 3.441,
+      "step": 75200
+    },
+    {
+      "epoch": 0.753,
+      "grad_norm": 0.08924631774425507,
+      "learning_rate": 0.00016297369409447486,
+      "loss": 3.4413,
+      "step": 75300
+    },
+    {
+      "epoch": 0.754,
+      "grad_norm": 0.09695728868246078,
+      "learning_rate": 0.00016286558549611407,
+      "loss": 3.4417,
+      "step": 75400
+    },
+    {
+      "epoch": 0.755,
+      "grad_norm": 0.08458244055509567,
+      "learning_rate": 0.00016275769175423189,
+      "loss": 3.4414,
+      "step": 75500
+    },
+    {
+      "epoch": 0.756,
+      "grad_norm": 0.09474821388721466,
+      "learning_rate": 0.00016265001215808885,
+      "loss": 3.4387,
+      "step": 75600
+    },
+    {
+      "epoch": 0.757,
+      "grad_norm": 0.0918809249997139,
+      "learning_rate": 0.00016254254600023274,
+      "loss": 3.4394,
+      "step": 75700
+    },
+    {
+      "epoch": 0.758,
+      "grad_norm": 0.10976070165634155,
+      "learning_rate": 0.000162435292576479,
+      "loss": 3.4403,
+      "step": 75800
+    },
+    {
+      "epoch": 0.759,
+      "grad_norm": 0.1117151528596878,
+      "learning_rate": 0.00016232825118589134,
+      "loss": 3.4399,
+      "step": 75900
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.12362287938594818,
+      "learning_rate": 0.00016222142113076255,
+      "loss": 3.4402,
+      "step": 76000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 3.8828125,
+      "eval_runtime": 32.9906,
+      "eval_samples_per_second": 18253.325,
+      "eval_steps_per_second": 17.854,
+      "step": 76000
+    },
+    {
+      "epoch": 0.761,
+      "grad_norm": 0.08545960485935211,
+      "learning_rate": 0.0001621148017165954,
+      "loss": 3.4404,
+      "step": 76100
+    },
+    {
+      "epoch": 0.762,
+      "grad_norm": 0.11282055079936981,
+      "learning_rate": 0.00016200839225208365,
+      "loss": 3.4406,
+      "step": 76200
+    },
+    {
+      "epoch": 0.763,
+      "grad_norm": 0.1176215335726738,
+      "learning_rate": 0.00016190219204909316,
+      "loss": 3.4415,
+      "step": 76300
+    },
+    {
+      "epoch": 0.764,
+      "grad_norm": 0.09560921788215637,
+      "learning_rate": 0.0001617962004226434,
+      "loss": 3.4413,
+      "step": 76400
+    },
+    {
+      "epoch": 0.765,
+      "grad_norm": 0.09105270355939865,
+      "learning_rate": 0.00016169041669088867,
+      "loss": 3.4409,
+      "step": 76500
+    },
+    {
+      "epoch": 0.766,
+      "grad_norm": 0.0954379066824913,
+      "learning_rate": 0.00016158484017509978,
+      "loss": 3.4382,
+      "step": 76600
+    },
+    {
+      "epoch": 0.767,
+      "grad_norm": 0.09138867259025574,
+      "learning_rate": 0.00016147947019964576,
+      "loss": 3.4412,
+      "step": 76700
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 0.09806874394416809,
+      "learning_rate": 0.0001613743060919757,
+      "loss": 3.442,
+      "step": 76800
+    },
+    {
+      "epoch": 0.769,
+      "grad_norm": 0.08680954575538635,
+      "learning_rate": 0.00016126934718260072,
+      "loss": 3.4406,
+      "step": 76900
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.10539354383945465,
+      "learning_rate": 0.00016116459280507607,
+      "loss": 3.4412,
+      "step": 77000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 3.88671875,
+      "eval_runtime": 33.0031,
+      "eval_samples_per_second": 18246.399,
+      "eval_steps_per_second": 17.847,
+      "step": 77000
+    },
+    {
+      "epoch": 0.771,
+      "grad_norm": 0.09596407413482666,
+      "learning_rate": 0.0001610600422959833,
+      "loss": 3.439,
+      "step": 77100
+    },
+    {
+      "epoch": 0.772,
+      "grad_norm": 0.09768424183130264,
+      "learning_rate": 0.00016095569499491262,
+      "loss": 3.4379,
+      "step": 77200
+    },
+    {
+      "epoch": 0.773,
+      "grad_norm": 0.09115412831306458,
+      "learning_rate": 0.0001608515502444456,
+      "loss": 3.4381,
+      "step": 77300
+    },
+    {
+      "epoch": 0.774,
+      "grad_norm": 0.11213884502649307,
+      "learning_rate": 0.00016074760739013738,
+      "loss": 3.4402,
+      "step": 77400
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 0.08757272362709045,
+      "learning_rate": 0.00016064386578049978,
+      "loss": 3.438,
+      "step": 77500
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 0.08894165605306625,
+      "learning_rate": 0.0001605403247669839,
+      "loss": 3.4417,
+      "step": 77600
+    },
+    {
+      "epoch": 0.777,
+      "grad_norm": 0.10517994314432144,
+      "learning_rate": 0.00016043698370396314,
+      "loss": 3.4397,
+      "step": 77700
+    },
+    {
+      "epoch": 0.778,
+      "grad_norm": 0.08438605815172195,
+      "learning_rate": 0.00016033384194871647,
+      "loss": 3.4374,
+      "step": 77800
+    },
+    {
+      "epoch": 0.779,
+      "grad_norm": 0.1192629337310791,
+      "learning_rate": 0.00016023089886141128,
+      "loss": 3.4374,
+      "step": 77900
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.11247923970222473,
+      "learning_rate": 0.00016012815380508712,
+      "loss": 3.438,
+      "step": 78000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 3.875,
+      "eval_runtime": 32.9875,
+      "eval_samples_per_second": 18255.029,
+      "eval_steps_per_second": 17.855,
+      "step": 78000
+    },
+    {
+      "epoch": 0.781,
+      "grad_norm": 0.1209215372800827,
+      "learning_rate": 0.0001600256061456389,
+      "loss": 3.4371,
+      "step": 78100
+    },
+    {
+      "epoch": 0.782,
+      "grad_norm": 0.09779994189739227,
+      "learning_rate": 0.00015992325525180032,
+      "loss": 3.4398,
+      "step": 78200
+    },
+    {
+      "epoch": 0.783,
+      "grad_norm": 0.09656972438097,
+      "learning_rate": 0.00015982110049512805,
+      "loss": 3.4385,
+      "step": 78300
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 0.08916040509939194,
+      "learning_rate": 0.000159719141249985,
+      "loss": 3.4366,
+      "step": 78400
+    },
+    {
+      "epoch": 0.785,
+      "grad_norm": 0.09170043468475342,
+      "learning_rate": 0.00015961737689352442,
+      "loss": 3.438,
+      "step": 78500
+    },
+    {
+      "epoch": 0.786,
+      "grad_norm": 0.09335237741470337,
+      "learning_rate": 0.0001595158068056741,
+      "loss": 3.438,
+      "step": 78600
+    },
+    {
+      "epoch": 0.787,
+      "grad_norm": 0.09777306765317917,
+      "learning_rate": 0.00015941443036912014,
+      "loss": 3.4364,
+      "step": 78700
+    },
+    {
+      "epoch": 0.788,
+      "grad_norm": 0.08855511248111725,
+      "learning_rate": 0.00015931324696929155,
+      "loss": 3.4368,
+      "step": 78800
+    },
+    {
+      "epoch": 0.789,
+      "grad_norm": 0.0980067327618599,
+      "learning_rate": 0.00015921225599434429,
+      "loss": 3.4375,
+      "step": 78900
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.09301699697971344,
+      "learning_rate": 0.000159111456835146,
+      "loss": 3.4359,
+      "step": 79000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 3.875,
+      "eval_runtime": 33.0055,
+      "eval_samples_per_second": 18245.074,
+      "eval_steps_per_second": 17.846,
+      "step": 79000
+    },
+    {
+      "epoch": 0.791,
+      "grad_norm": 0.10222512483596802,
+      "learning_rate": 0.00015901084888526044,
+      "loss": 3.4359,
+      "step": 79100
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 0.09246222674846649,
+      "learning_rate": 0.00015891043154093204,
+      "loss": 3.4358,
+      "step": 79200
+    },
+    {
+      "epoch": 0.793,
+      "grad_norm": 0.11724498867988586,
+      "learning_rate": 0.00015881020420107104,
+      "loss": 3.437,
+      "step": 79300
+    },
+    {
+      "epoch": 0.794,
+      "grad_norm": 0.08812806755304337,
+      "learning_rate": 0.0001587101662672379,
+      "loss": 3.4366,
+      "step": 79400
+    },
+    {
+      "epoch": 0.795,
+      "grad_norm": 0.11343426257371902,
+      "learning_rate": 0.00015861031714362882,
+      "loss": 3.4361,
+      "step": 79500
+    },
+    {
+      "epoch": 0.796,
+      "grad_norm": 0.08835262805223465,
+      "learning_rate": 0.00015851065623706038,
+      "loss": 3.4358,
+      "step": 79600
+    },
+    {
+      "epoch": 0.797,
+      "grad_norm": 0.08559754490852356,
+      "learning_rate": 0.00015841118295695486,
+      "loss": 3.4342,
+      "step": 79700
+    },
+    {
+      "epoch": 0.798,
+      "grad_norm": 0.13493484258651733,
+      "learning_rate": 0.0001583118967153259,
+      "loss": 3.4362,
+      "step": 79800
+    },
+    {
+      "epoch": 0.799,
+      "grad_norm": 0.08687994629144669,
+      "learning_rate": 0.00015821279692676326,
+      "loss": 3.4368,
+      "step": 79900
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.08848557621240616,
+      "learning_rate": 0.00015811388300841897,
+      "loss": 3.4343,
+      "step": 80000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 3.869140625,
+      "eval_runtime": 32.9882,
+      "eval_samples_per_second": 18254.673,
+      "eval_steps_per_second": 17.855,
+      "step": 80000
+    },
+    {
+      "epoch": 0.801,
+      "grad_norm": 0.08878805488348007,
+      "learning_rate": 0.00015801515437999242,
+      "loss": 3.4347,
+      "step": 80100
+    },
+    {
+      "epoch": 0.802,
+      "grad_norm": 0.09379617869853973,
+      "learning_rate": 0.00015791661046371636,
+      "loss": 3.4347,
+      "step": 80200
+    },
+    {
+      "epoch": 0.803,
+      "grad_norm": 0.08302991092205048,
+      "learning_rate": 0.00015781825068434261,
+      "loss": 3.4346,
+      "step": 80300
+    },
+    {
+      "epoch": 0.804,
+      "grad_norm": 0.10080291330814362,
+      "learning_rate": 0.00015772007446912791,
+      "loss": 3.4362,
+      "step": 80400
+    },
+    {
+      "epoch": 0.805,
+      "grad_norm": 0.1289501041173935,
+      "learning_rate": 0.00015762208124782013,
+      "loss": 3.4357,
+      "step": 80500
+    },
+    {
+      "epoch": 0.806,
+      "grad_norm": 0.0920480489730835,
+      "learning_rate": 0.00015752427045264398,
+      "loss": 3.4349,
+      "step": 80600
+    },
+    {
+      "epoch": 0.807,
+      "grad_norm": 0.09406436234712601,
+      "learning_rate": 0.00015742664151828745,
+      "loss": 3.4365,
+      "step": 80700
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 0.09713851660490036,
+      "learning_rate": 0.00015732919388188816,
+      "loss": 3.4365,
+      "step": 80800
+    },
+    {
+      "epoch": 0.809,
+      "grad_norm": 0.09240105003118515,
+      "learning_rate": 0.0001572319269830195,
+      "loss": 3.4357,
+      "step": 80900
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.09189952164888382,
+      "learning_rate": 0.00015713484026367723,
+      "loss": 3.4325,
+      "step": 81000
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 3.87109375,
+      "eval_runtime": 33.0008,
+      "eval_samples_per_second": 18247.706,
+      "eval_steps_per_second": 17.848,
+      "step": 81000
+    },
+    {
+      "epoch": 0.811,
+      "grad_norm": 0.09463001787662506,
+      "learning_rate": 0.00015703793316826606,
+      "loss": 3.4342,
+      "step": 81100
+    },
+    {
+      "epoch": 0.812,
+      "grad_norm": 0.11351007223129272,
+      "learning_rate": 0.0001569412051435861,
+      "loss": 3.4349,
+      "step": 81200
+    },
+    {
+      "epoch": 0.813,
+      "grad_norm": 0.09494343400001526,
+      "learning_rate": 0.00015684465563882,
+      "loss": 3.435,
+      "step": 81300
+    },
+    {
+      "epoch": 0.814,
+      "grad_norm": 0.08718815445899963,
+      "learning_rate": 0.00015674828410551923,
+      "loss": 3.4344,
+      "step": 81400
+    },
+    {
+      "epoch": 0.815,
+      "grad_norm": 0.08435869216918945,
+      "learning_rate": 0.00015665208999759148,
+      "loss": 3.4326,
+      "step": 81500
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 0.11424301564693451,
+      "learning_rate": 0.0001565560727712874,
+      "loss": 3.4322,
+      "step": 81600
+    },
+    {
+      "epoch": 0.817,
+      "grad_norm": 0.11047474294900894,
+      "learning_rate": 0.0001564602318851877,
+      "loss": 3.4343,
+      "step": 81700
+    },
+    {
+      "epoch": 0.818,
+      "grad_norm": 0.09242834150791168,
+      "learning_rate": 0.00015636456680019053,
+      "loss": 3.4329,
+      "step": 81800
+    },
+    {
+      "epoch": 0.819,
+      "grad_norm": 0.10262425988912582,
+      "learning_rate": 0.00015626907697949846,
+      "loss": 3.4338,
+      "step": 81900
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.0893552154302597,
+      "learning_rate": 0.00015617376188860606,
+      "loss": 3.4346,
+      "step": 82000
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 3.876953125,
+      "eval_runtime": 32.9906,
+      "eval_samples_per_second": 18253.339,
+      "eval_steps_per_second": 17.854,
+      "step": 82000
+    },
+    {
+      "epoch": 0.821,
+      "grad_norm": 0.0904051810503006,
+      "learning_rate": 0.00015607862099528718,
+      "loss": 3.433,
+      "step": 82100
+    },
+    {
+      "epoch": 0.822,
+      "grad_norm": 0.08803316950798035,
+      "learning_rate": 0.00015598365376958255,
+      "loss": 3.431,
+      "step": 82200
+    },
+    {
+      "epoch": 0.823,
+      "grad_norm": 0.08486661314964294,
+      "learning_rate": 0.00015588885968378734,
+      "loss": 3.4327,
+      "step": 82300
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 0.1493000090122223,
+      "learning_rate": 0.00015579423821243896,
+      "loss": 3.4336,
+      "step": 82400
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 0.08111909031867981,
+      "learning_rate": 0.0001556997888323046,
+      "loss": 3.4309,
+      "step": 82500
+    },
+    {
+      "epoch": 0.826,
+      "grad_norm": 0.15636824071407318,
+      "learning_rate": 0.0001556055110223693,
+      "loss": 3.433,
+      "step": 82600
+    },
+    {
+      "epoch": 0.827,
+      "grad_norm": 0.09289142489433289,
+      "learning_rate": 0.00015551140426382372,
+      "loss": 3.4333,
+      "step": 82700
+    },
+    {
+      "epoch": 0.828,
+      "grad_norm": 0.10629736632108688,
+      "learning_rate": 0.0001554174680400523,
+      "loss": 3.432,
+      "step": 82800
+    },
+    {
+      "epoch": 0.829,
+      "grad_norm": 0.11100473254919052,
+      "learning_rate": 0.0001553237018366212,
+      "loss": 3.433,
+      "step": 82900
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.09321753680706024,
+      "learning_rate": 0.00015523010514126656,
+      "loss": 3.4343,
+      "step": 83000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 3.87109375,
+      "eval_runtime": 32.9766,
+      "eval_samples_per_second": 18261.078,
+      "eval_steps_per_second": 17.861,
+      "step": 83000
+    },
+    {
+      "epoch": 0.831,
+      "grad_norm": 0.1051269918680191,
+      "learning_rate": 0.00015513667744388273,
+      "loss": 3.4318,
+      "step": 83100
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 0.108842633664608,
+      "learning_rate": 0.00015504341823651055,
+      "loss": 3.4324,
+      "step": 83200
+    },
+    {
+      "epoch": 0.833,
+      "grad_norm": 0.11733662337064743,
+      "learning_rate": 0.00015495032701332584,
+      "loss": 3.434,
+      "step": 83300
+    },
+    {
+      "epoch": 0.834,
+      "grad_norm": 0.12677276134490967,
+      "learning_rate": 0.00015485740327062772,
+      "loss": 3.4328,
+      "step": 83400
+    },
+    {
+      "epoch": 0.835,
+      "grad_norm": 0.08611226826906204,
+      "learning_rate": 0.00015476464650682737,
+      "loss": 3.4333,
+      "step": 83500
+    },
+    {
+      "epoch": 0.836,
+      "grad_norm": 0.08939053118228912,
+      "learning_rate": 0.0001546720562224365,
+      "loss": 3.4316,
+      "step": 83600
+    },
+    {
+      "epoch": 0.837,
+      "grad_norm": 0.10289647430181503,
+      "learning_rate": 0.0001545796319200561,
+      "loss": 3.4311,
+      "step": 83700
+    },
+    {
+      "epoch": 0.838,
+      "grad_norm": 0.0950452983379364,
+      "learning_rate": 0.00015448737310436522,
+      "loss": 3.4332,
+      "step": 83800
+    },
+    {
+      "epoch": 0.839,
+      "grad_norm": 0.09242187440395355,
+      "learning_rate": 0.00015439527928210988,
+      "loss": 3.4311,
+      "step": 83900
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.10447569936513901,
+      "learning_rate": 0.00015430334996209192,
+      "loss": 3.4316,
+      "step": 84000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 3.875,
+      "eval_runtime": 32.9725,
+      "eval_samples_per_second": 18263.359,
+      "eval_steps_per_second": 17.863,
+      "step": 84000
+    },
+    {
+      "epoch": 0.841,
+      "grad_norm": 0.09993547201156616,
+      "learning_rate": 0.0001542115846551579,
+      "loss": 3.4326,
+      "step": 84100
+    },
+    {
+      "epoch": 0.842,
+      "grad_norm": 0.0941188782453537,
+      "learning_rate": 0.00015411998287418846,
+      "loss": 3.4318,
+      "step": 84200
+    },
+    {
+      "epoch": 0.843,
+      "grad_norm": 0.12041068077087402,
+      "learning_rate": 0.00015402854413408716,
+      "loss": 3.4319,
+      "step": 84300
+    },
+    {
+      "epoch": 0.844,
+      "grad_norm": 0.09113665670156479,
+      "learning_rate": 0.00015393726795176978,
+      "loss": 3.4323,
+      "step": 84400
+    },
+    {
+      "epoch": 0.845,
+      "grad_norm": 0.11387240141630173,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 3.4301,
+      "step": 84500
+    },
+    {
+      "epoch": 0.846,
+      "grad_norm": 0.09158744663000107,
+      "learning_rate": 0.0001537552013381475,
+      "loss": 3.4298,
+      "step": 84600
+    },
+    {
+      "epoch": 0.847,
+      "grad_norm": 0.08281726390123367,
+      "learning_rate": 0.00015366440995063938,
+      "loss": 3.4319,
+      "step": 84700
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 0.08520444482564926,
+      "learning_rate": 0.0001535737792084878,
+      "loss": 3.4314,
+      "step": 84800
+    },
+    {
+      "epoch": 0.849,
+      "grad_norm": 0.08099710196256638,
+      "learning_rate": 0.0001534833086385105,
+      "loss": 3.4322,
+      "step": 84900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.08628775179386139,
+      "learning_rate": 0.00015339299776947408,
+      "loss": 3.4306,
+      "step": 85000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 3.873046875,
+      "eval_runtime": 32.9817,
+      "eval_samples_per_second": 18258.237,
+      "eval_steps_per_second": 17.858,
+      "step": 85000
+    },
+    {
+      "epoch": 0.851,
+      "grad_norm": 0.08802857249975204,
+      "learning_rate": 0.00015330284613208398,
+      "loss": 3.4298,
+      "step": 85100
+    },
+    {
+      "epoch": 0.852,
+      "grad_norm": 0.09676956385374069,
+      "learning_rate": 0.00015321285325897388,
+      "loss": 3.4307,
+      "step": 85200
+    },
+    {
+      "epoch": 0.853,
+      "grad_norm": 0.10504265874624252,
+      "learning_rate": 0.00015312301868469587,
+      "loss": 3.4301,
+      "step": 85300
+    },
+    {
+      "epoch": 0.854,
+      "grad_norm": 0.08877792209386826,
+      "learning_rate": 0.00015303334194570998,
+      "loss": 3.4306,
+      "step": 85400
+    },
+    {
+      "epoch": 0.855,
+      "grad_norm": 0.12948188185691833,
+      "learning_rate": 0.0001529438225803745,
+      "loss": 3.4299,
+      "step": 85500
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 0.08437684923410416,
+      "learning_rate": 0.00015285446012893578,
+      "loss": 3.4306,
+      "step": 85600
+    },
+    {
+      "epoch": 0.857,
+      "grad_norm": 0.08995580673217773,
+      "learning_rate": 0.00015276525413351826,
+      "loss": 3.4316,
+      "step": 85700
+    },
+    {
+      "epoch": 0.858,
+      "grad_norm": 0.08715008944272995,
+      "learning_rate": 0.00015267620413811483,
+      "loss": 3.4298,
+      "step": 85800
+    },
+    {
+      "epoch": 0.859,
+      "grad_norm": 0.0873521938920021,
+      "learning_rate": 0.0001525873096885769,
+      "loss": 3.428,
+      "step": 85900
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.08357945084571838,
+      "learning_rate": 0.00015249857033260467,
+      "loss": 3.4303,
+      "step": 86000
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 3.873046875,
+      "eval_runtime": 32.9692,
+      "eval_samples_per_second": 18265.15,
+      "eval_steps_per_second": 17.865,
+      "step": 86000
+    },
+    {
+      "epoch": 0.861,
+      "grad_norm": 0.0914229154586792,
+      "learning_rate": 0.00015240998561973752,
+      "loss": 3.4303,
+      "step": 86100
+    },
+    {
+      "epoch": 0.862,
+      "grad_norm": 0.10973150283098221,
+      "learning_rate": 0.00015232155510134433,
+      "loss": 3.4277,
+      "step": 86200
+    },
+    {
+      "epoch": 0.863,
+      "grad_norm": 0.1186969205737114,
+      "learning_rate": 0.00015223327833061405,
+      "loss": 3.4279,
+      "step": 86300
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 0.08916877210140228,
+      "learning_rate": 0.00015214515486254611,
+      "loss": 3.4301,
+      "step": 86400
+    },
+    {
+      "epoch": 0.865,
+      "grad_norm": 0.09595561027526855,
+      "learning_rate": 0.0001520571842539411,
+      "loss": 3.4272,
+      "step": 86500
+    },
+    {
+      "epoch": 0.866,
+      "grad_norm": 0.10776854306459427,
+      "learning_rate": 0.00015196936606339122,
+      "loss": 3.4285,
+      "step": 86600
+    },
+    {
+      "epoch": 0.867,
+      "grad_norm": 0.09018661081790924,
+      "learning_rate": 0.00015188169985127126,
+      "loss": 3.4284,
+      "step": 86700
+    },
+    {
+      "epoch": 0.868,
+      "grad_norm": 0.09676007181406021,
+      "learning_rate": 0.0001517941851797291,
+      "loss": 3.428,
+      "step": 86800
+    },
+    {
+      "epoch": 0.869,
+      "grad_norm": 0.09606127440929413,
+      "learning_rate": 0.0001517068216126766,
+      "loss": 3.4301,
+      "step": 86900
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0857662633061409,
+      "learning_rate": 0.0001516196087157807,
+      "loss": 3.431,
+      "step": 87000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 3.87109375,
+      "eval_runtime": 32.9775,
+      "eval_samples_per_second": 18260.568,
+      "eval_steps_per_second": 17.861,
+      "step": 87000
+    },
+    {
+      "epoch": 0.871,
+      "grad_norm": 0.11101428419351578,
+      "learning_rate": 0.0001515325460564539,
+      "loss": 3.4315,
+      "step": 87100
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 0.10998611152172089,
+      "learning_rate": 0.00015144563320384566,
+      "loss": 3.4312,
+      "step": 87200
+    },
+    {
+      "epoch": 0.873,
+      "grad_norm": 0.0887598916888237,
+      "learning_rate": 0.0001513588697288333,
+      "loss": 3.4287,
+      "step": 87300
+    },
+    {
+      "epoch": 0.874,
+      "grad_norm": 0.09869083762168884,
+      "learning_rate": 0.00015127225520401288,
+      "loss": 3.4286,
+      "step": 87400
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 0.09993433952331543,
+      "learning_rate": 0.0001511857892036909,
+      "loss": 3.4268,
+      "step": 87500
+    },
+    {
+      "epoch": 0.876,
+      "grad_norm": 0.09299583733081818,
+      "learning_rate": 0.00015109947130387486,
+      "loss": 3.4288,
+      "step": 87600
+    },
+    {
+      "epoch": 0.877,
+      "grad_norm": 0.09471836686134338,
+      "learning_rate": 0.00015101330108226502,
+      "loss": 3.4278,
+      "step": 87700
+    },
+    {
+      "epoch": 0.878,
+      "grad_norm": 0.10189653187990189,
+      "learning_rate": 0.00015092727811824553,
+      "loss": 3.4277,
+      "step": 87800
+    },
+    {
+      "epoch": 0.879,
+      "grad_norm": 0.09086822718381882,
+      "learning_rate": 0.00015084140199287576,
+      "loss": 3.4283,
+      "step": 87900
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.08800884336233139,
+      "learning_rate": 0.00015075567228888182,
+      "loss": 3.43,
+      "step": 88000
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 3.8671875,
+      "eval_runtime": 32.9826,
+      "eval_samples_per_second": 18257.767,
+      "eval_steps_per_second": 17.858,
+      "step": 88000
+    },
+    {
+      "epoch": 0.881,
+      "grad_norm": 0.08497955650091171,
+      "learning_rate": 0.00015067008859064807,
+      "loss": 3.4282,
+      "step": 88100
+    },
+    {
+      "epoch": 0.882,
+      "grad_norm": 0.09592936187982559,
+      "learning_rate": 0.00015058465048420854,
+      "loss": 3.4292,
+      "step": 88200
+    },
+    {
+      "epoch": 0.883,
+      "grad_norm": 0.08380814641714096,
+      "learning_rate": 0.00015049935755723866,
+      "loss": 3.4281,
+      "step": 88300
+    },
+    {
+      "epoch": 0.884,
+      "grad_norm": 0.08948108553886414,
+      "learning_rate": 0.0001504142093990467,
+      "loss": 3.4273,
+      "step": 88400
+    },
+    {
+      "epoch": 0.885,
+      "grad_norm": 0.09598647058010101,
+      "learning_rate": 0.00015032920560056578,
+      "loss": 3.4282,
+      "step": 88500
+    },
+    {
+      "epoch": 0.886,
+      "grad_norm": 0.09412466734647751,
+      "learning_rate": 0.00015024434575434525,
+      "loss": 3.4272,
+      "step": 88600
+    },
+    {
+      "epoch": 0.887,
+      "grad_norm": 0.08592375367879868,
+      "learning_rate": 0.0001501596294545428,
+      "loss": 3.4258,
+      "step": 88700
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 0.09520655870437622,
+      "learning_rate": 0.00015007505629691605,
+      "loss": 3.4295,
+      "step": 88800
+    },
+    {
+      "epoch": 0.889,
+      "grad_norm": 0.08593907207250595,
+      "learning_rate": 0.0001499906258788147,
+      "loss": 3.4263,
+      "step": 88900
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.09926886111497879,
+      "learning_rate": 0.00014990633779917229,
+      "loss": 3.4279,
+      "step": 89000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 3.86328125,
+      "eval_runtime": 32.9884,
+      "eval_samples_per_second": 18254.532,
+      "eval_steps_per_second": 17.855,
+      "step": 89000
+    },
+    {
+      "epoch": 0.891,
+      "grad_norm": 0.09873060137033463,
+      "learning_rate": 0.00014982219165849827,
+      "loss": 3.4272,
+      "step": 89100
+    },
+    {
+      "epoch": 0.892,
+      "grad_norm": 0.09635820984840393,
+      "learning_rate": 0.00014973818705886998,
+      "loss": 3.4278,
+      "step": 89200
+    },
+    {
+      "epoch": 0.893,
+      "grad_norm": 0.08815903961658478,
+      "learning_rate": 0.00014965432360392487,
+      "loss": 3.4278,
+      "step": 89300
+    },
+    {
+      "epoch": 0.894,
+      "grad_norm": 0.08887197822332382,
+      "learning_rate": 0.00014957060089885261,
+      "loss": 3.4264,
+      "step": 89400
+    },
+    {
+      "epoch": 0.895,
+      "grad_norm": 0.08929969370365143,
+      "learning_rate": 0.00014948701855038718,
+      "loss": 3.4282,
+      "step": 89500
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 0.09338970482349396,
+      "learning_rate": 0.00014940357616679919,
+      "loss": 3.4288,
+      "step": 89600
+    },
+    {
+      "epoch": 0.897,
+      "grad_norm": 0.1054818257689476,
+      "learning_rate": 0.00014932027335788826,
+      "loss": 3.4285,
+      "step": 89700
+    },
+    {
+      "epoch": 0.898,
+      "grad_norm": 0.09273470193147659,
+      "learning_rate": 0.0001492371097349751,
+      "loss": 3.4276,
+      "step": 89800
+    },
+    {
+      "epoch": 0.899,
+      "grad_norm": 0.08965855836868286,
+      "learning_rate": 0.00014915408491089415,
+      "loss": 3.4274,
+      "step": 89900
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.11126144975423813,
+      "learning_rate": 0.00014907119849998598,
+      "loss": 3.4265,
+      "step": 90000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 3.87109375,
+      "eval_runtime": 32.9878,
+      "eval_samples_per_second": 18254.881,
+      "eval_steps_per_second": 17.855,
+      "step": 90000
+    },
+    {
+      "epoch": 0.901,
+      "grad_norm": 0.09338975697755814,
+      "learning_rate": 0.00014898845011808955,
+      "loss": 3.4256,
+      "step": 90100
+    },
+    {
+      "epoch": 0.902,
+      "grad_norm": 0.09844280779361725,
+      "learning_rate": 0.00014890583938253493,
+      "loss": 3.4274,
+      "step": 90200
+    },
+    {
+      "epoch": 0.903,
+      "grad_norm": 0.08697222173213959,
+      "learning_rate": 0.0001488233659121359,
+      "loss": 3.4254,
+      "step": 90300
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 0.09345566481351852,
+      "learning_rate": 0.0001487410293271824,
+      "loss": 3.4268,
+      "step": 90400
+    },
+    {
+      "epoch": 0.905,
+      "grad_norm": 0.08729152381420135,
+      "learning_rate": 0.00014865882924943328,
+      "loss": 3.4251,
+      "step": 90500
+    },
+    {
+      "epoch": 0.906,
+      "grad_norm": 0.12593166530132294,
+      "learning_rate": 0.00014857676530210894,
+      "loss": 3.4275,
+      "step": 90600
+    },
+    {
+      "epoch": 0.907,
+      "grad_norm": 0.0890730693936348,
+      "learning_rate": 0.00014849483710988428,
+      "loss": 3.4251,
+      "step": 90700
+    },
+    {
+      "epoch": 0.908,
+      "grad_norm": 0.08751527220010757,
+      "learning_rate": 0.0001484130442988812,
+      "loss": 3.4244,
+      "step": 90800
+    },
+    {
+      "epoch": 0.909,
+      "grad_norm": 0.11215963959693909,
+      "learning_rate": 0.00014833138649666158,
+      "loss": 3.424,
+      "step": 90900
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.101831816136837,
+      "learning_rate": 0.00014824986333222023,
+      "loss": 3.4266,
+      "step": 91000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 3.857421875,
+      "eval_runtime": 32.9906,
+      "eval_samples_per_second": 18253.323,
+      "eval_steps_per_second": 17.854,
+      "step": 91000
+    },
+    {
+      "epoch": 0.911,
+      "grad_norm": 0.09032022207975388,
+      "learning_rate": 0.00014816847443597765,
+      "loss": 3.4258,
+      "step": 91100
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 0.11087517440319061,
+      "learning_rate": 0.00014808721943977308,
+      "loss": 3.4257,
+      "step": 91200
+    },
+    {
+      "epoch": 0.913,
+      "grad_norm": 0.08949613571166992,
+      "learning_rate": 0.00014800609797685756,
+      "loss": 3.4256,
+      "step": 91300
+    },
+    {
+      "epoch": 0.914,
+      "grad_norm": 0.08762556314468384,
+      "learning_rate": 0.00014792510968188681,
+      "loss": 3.4229,
+      "step": 91400
+    },
+    {
+      "epoch": 0.915,
+      "grad_norm": 0.09100620448589325,
+      "learning_rate": 0.00014784425419091458,
+      "loss": 3.4263,
+      "step": 91500
+    },
+    {
+      "epoch": 0.916,
+      "grad_norm": 0.10705144703388214,
+      "learning_rate": 0.00014776353114138543,
+      "loss": 3.4235,
+      "step": 91600
+    },
+    {
+      "epoch": 0.917,
+      "grad_norm": 0.1440095603466034,
+      "learning_rate": 0.00014768294017212822,
+      "loss": 3.4259,
+      "step": 91700
+    },
+    {
+      "epoch": 0.918,
+      "grad_norm": 0.11515694111585617,
+      "learning_rate": 0.00014760248092334923,
+      "loss": 3.4233,
+      "step": 91800
+    },
+    {
+      "epoch": 0.919,
+      "grad_norm": 0.08891148865222931,
+      "learning_rate": 0.00014752215303662526,
+      "loss": 3.4228,
+      "step": 91900
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.0852489098906517,
+      "learning_rate": 0.00014744195615489715,
+      "loss": 3.4253,
+      "step": 92000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 3.869140625,
+      "eval_runtime": 32.971,
+      "eval_samples_per_second": 18264.179,
+      "eval_steps_per_second": 17.864,
+      "step": 92000
+    },
+    {
+      "epoch": 0.921,
+      "grad_norm": 0.09289970248937607,
+      "learning_rate": 0.00014736188992246294,
+      "loss": 3.4251,
+      "step": 92100
+    },
+    {
+      "epoch": 0.922,
+      "grad_norm": 0.09001227468252182,
+      "learning_rate": 0.0001472819539849714,
+      "loss": 3.4238,
+      "step": 92200
+    },
+    {
+      "epoch": 0.923,
+      "grad_norm": 0.08918751776218414,
+      "learning_rate": 0.0001472021479894153,
+      "loss": 3.4254,
+      "step": 92300
+    },
+    {
+      "epoch": 0.924,
+      "grad_norm": 0.11080797016620636,
+      "learning_rate": 0.00014712247158412493,
+      "loss": 3.4238,
+      "step": 92400
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 0.08473057299852371,
+      "learning_rate": 0.00014704292441876154,
+      "loss": 3.426,
+      "step": 92500
+    },
+    {
+      "epoch": 0.926,
+      "grad_norm": 0.11121894419193268,
+      "learning_rate": 0.00014696350614431104,
+      "loss": 3.4233,
+      "step": 92600
+    },
+    {
+      "epoch": 0.927,
+      "grad_norm": 0.0871448889374733,
+      "learning_rate": 0.00014688421641307725,
+      "loss": 3.4233,
+      "step": 92700
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 0.0890325978398323,
+      "learning_rate": 0.00014680505487867586,
+      "loss": 3.4248,
+      "step": 92800
+    },
+    {
+      "epoch": 0.929,
+      "grad_norm": 0.10418891906738281,
+      "learning_rate": 0.0001467260211960279,
+      "loss": 3.4221,
+      "step": 92900
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.0831473171710968,
+      "learning_rate": 0.0001466471150213533,
+      "loss": 3.4226,
+      "step": 93000
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 3.869140625,
+      "eval_runtime": 32.9976,
+      "eval_samples_per_second": 18249.449,
+      "eval_steps_per_second": 17.85,
+      "step": 93000
+    },
+    {
+      "epoch": 0.931,
+      "grad_norm": 0.0834508016705513,
+      "learning_rate": 0.00014656833601216487,
+      "loss": 3.4233,
+      "step": 93100
+    },
+    {
+      "epoch": 0.932,
+      "grad_norm": 0.0903562530875206,
+      "learning_rate": 0.0001464896838272619,
+      "loss": 3.424,
+      "step": 93200
+    },
+    {
+      "epoch": 0.933,
+      "grad_norm": 0.1055062785744667,
+      "learning_rate": 0.00014641115812672397,
+      "loss": 3.4246,
+      "step": 93300
+    },
+    {
+      "epoch": 0.934,
+      "grad_norm": 0.08662033081054688,
+      "learning_rate": 0.00014633275857190483,
+      "loss": 3.4235,
+      "step": 93400
+    },
+    {
+      "epoch": 0.935,
+      "grad_norm": 0.08579885214567184,
+      "learning_rate": 0.00014625448482542613,
+      "loss": 3.425,
+      "step": 93500
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 0.09081684798002243,
+      "learning_rate": 0.00014617633655117154,
+      "loss": 3.4235,
+      "step": 93600
+    },
+    {
+      "epoch": 0.937,
+      "grad_norm": 0.11446008086204529,
+      "learning_rate": 0.00014609831341428047,
+      "loss": 3.4207,
+      "step": 93700
+    },
+    {
+      "epoch": 0.938,
+      "grad_norm": 0.09146326035261154,
+      "learning_rate": 0.00014602041508114227,
+      "loss": 3.4223,
+      "step": 93800
+    },
+    {
+      "epoch": 0.939,
+      "grad_norm": 0.09234822541475296,
+      "learning_rate": 0.00014594264121938996,
+      "loss": 3.4204,
+      "step": 93900
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.09005927294492722,
+      "learning_rate": 0.00014586499149789455,
+      "loss": 3.4208,
+      "step": 94000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 3.859375,
+      "eval_runtime": 33.0129,
+      "eval_samples_per_second": 18240.999,
+      "eval_steps_per_second": 17.842,
+      "step": 94000
+    },
+    {
+      "epoch": 0.941,
+      "grad_norm": 0.09933404624462128,
+      "learning_rate": 0.00014578746558675893,
+      "loss": 3.4215,
+      "step": 94100
+    },
+    {
+      "epoch": 0.942,
+      "grad_norm": 0.12064062803983688,
+      "learning_rate": 0.000145710063157312,
+      "loss": 3.4215,
+      "step": 94200
+    },
+    {
+      "epoch": 0.943,
+      "grad_norm": 0.09077318012714386,
+      "learning_rate": 0.00014563278388210303,
+      "loss": 3.4224,
+      "step": 94300
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 0.10471939295530319,
+      "learning_rate": 0.0001455556274348955,
+      "loss": 3.4217,
+      "step": 94400
+    },
+    {
+      "epoch": 0.945,
+      "grad_norm": 0.11697626858949661,
+      "learning_rate": 0.00014547859349066158,
+      "loss": 3.4203,
+      "step": 94500
+    },
+    {
+      "epoch": 0.946,
+      "grad_norm": 0.10920999944210052,
+      "learning_rate": 0.00014540168172557633,
+      "loss": 3.4236,
+      "step": 94600
+    },
+    {
+      "epoch": 0.947,
+      "grad_norm": 0.10700856149196625,
+      "learning_rate": 0.0001453248918170118,
+      "loss": 3.4234,
+      "step": 94700
+    },
+    {
+      "epoch": 0.948,
+      "grad_norm": 0.0889246016740799,
+      "learning_rate": 0.0001452482234435317,
+      "loss": 3.4204,
+      "step": 94800
+    },
+    {
+      "epoch": 0.949,
+      "grad_norm": 0.09357952326536179,
+      "learning_rate": 0.00014517167628488533,
+      "loss": 3.4208,
+      "step": 94900
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.09217779338359833,
+      "learning_rate": 0.00014509525002200232,
+      "loss": 3.4228,
+      "step": 95000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 3.865234375,
+      "eval_runtime": 32.9993,
+      "eval_samples_per_second": 18248.486,
+      "eval_steps_per_second": 17.849,
+      "step": 95000
+    },
+    {
+      "epoch": 0.951,
+      "grad_norm": 0.08662614971399307,
+      "learning_rate": 0.00014501894433698685,
+      "loss": 3.4219,
+      "step": 95100
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 0.10022010654211044,
+      "learning_rate": 0.00014494275891311213,
+      "loss": 3.4205,
+      "step": 95200
+    },
+    {
+      "epoch": 0.953,
+      "grad_norm": 0.09261790663003922,
+      "learning_rate": 0.00014486669343481484,
+      "loss": 3.4234,
+      "step": 95300
+    },
+    {
+      "epoch": 0.954,
+      "grad_norm": 0.0873044952750206,
+      "learning_rate": 0.00014479074758768978,
+      "loss": 3.4213,
+      "step": 95400
+    },
+    {
+      "epoch": 0.955,
+      "grad_norm": 0.09158512949943542,
+      "learning_rate": 0.00014471492105848434,
+      "loss": 3.4232,
+      "step": 95500
+    },
+    {
+      "epoch": 0.956,
+      "grad_norm": 0.1242978498339653,
+      "learning_rate": 0.00014463921353509294,
+      "loss": 3.4225,
+      "step": 95600
+    },
+    {
+      "epoch": 0.957,
+      "grad_norm": 0.09228156507015228,
+      "learning_rate": 0.00014456362470655182,
+      "loss": 3.4216,
+      "step": 95700
+    },
+    {
+      "epoch": 0.958,
+      "grad_norm": 0.11788316071033478,
+      "learning_rate": 0.00014448815426303362,
+      "loss": 3.4207,
+      "step": 95800
+    },
+    {
+      "epoch": 0.959,
+      "grad_norm": 0.08355174213647842,
+      "learning_rate": 0.00014441280189584203,
+      "loss": 3.4211,
+      "step": 95900
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1150907650589943,
+      "learning_rate": 0.00014433756729740648,
+      "loss": 3.421,
+      "step": 96000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 3.86328125,
+      "eval_runtime": 32.9989,
+      "eval_samples_per_second": 18248.745,
+      "eval_steps_per_second": 17.849,
+      "step": 96000
+    },
+    {
+      "epoch": 0.961,
+      "grad_norm": 0.09715475887060165,
+      "learning_rate": 0.00014426245016127675,
+      "loss": 3.421,
+      "step": 96100
+    },
+    {
+      "epoch": 0.962,
+      "grad_norm": 0.08516929298639297,
+      "learning_rate": 0.00014418745018211809,
+      "loss": 3.4204,
+      "step": 96200
+    },
+    {
+      "epoch": 0.963,
+      "grad_norm": 0.09556137025356293,
+      "learning_rate": 0.00014411256705570565,
+      "loss": 3.4213,
+      "step": 96300
+    },
+    {
+      "epoch": 0.964,
+      "grad_norm": 0.11192219704389572,
+      "learning_rate": 0.00014403780047891933,
+      "loss": 3.42,
+      "step": 96400
+    },
+    {
+      "epoch": 0.965,
+      "grad_norm": 0.12303122878074646,
+      "learning_rate": 0.0001439631501497389,
+      "loss": 3.4192,
+      "step": 96500
+    },
+    {
+      "epoch": 0.966,
+      "grad_norm": 0.08821086585521698,
+      "learning_rate": 0.00014388861576723855,
+      "loss": 3.4192,
+      "step": 96600
+    },
+    {
+      "epoch": 0.967,
+      "grad_norm": 0.10734008997678757,
+      "learning_rate": 0.00014381419703158197,
+      "loss": 3.4211,
+      "step": 96700
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 0.08126262575387955,
+      "learning_rate": 0.00014373989364401728,
+      "loss": 3.4206,
+      "step": 96800
+    },
+    {
+      "epoch": 0.969,
+      "grad_norm": 0.10735704749822617,
+      "learning_rate": 0.00014366570530687184,
+      "loss": 3.4211,
+      "step": 96900
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.08785036206245422,
+      "learning_rate": 0.00014359163172354765,
+      "loss": 3.4198,
+      "step": 97000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 3.861328125,
+      "eval_runtime": 32.9832,
+      "eval_samples_per_second": 18257.433,
+      "eval_steps_per_second": 17.858,
+      "step": 97000
+    },
+    {
+      "epoch": 0.971,
+      "grad_norm": 0.0964650809764862,
+      "learning_rate": 0.00014351767259851572,
+      "loss": 3.42,
+      "step": 97100
+    },
+    {
+      "epoch": 0.972,
+      "grad_norm": 0.09006300568580627,
+      "learning_rate": 0.00014344382763731173,
+      "loss": 3.4206,
+      "step": 97200
+    },
+    {
+      "epoch": 0.973,
+      "grad_norm": 0.11106297373771667,
+      "learning_rate": 0.0001433700965465308,
+      "loss": 3.4203,
+      "step": 97300
+    },
+    {
+      "epoch": 0.974,
+      "grad_norm": 0.08400940150022507,
+      "learning_rate": 0.0001432964790338226,
+      "loss": 3.421,
+      "step": 97400
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 0.08784104138612747,
+      "learning_rate": 0.0001432229748078866,
+      "loss": 3.4188,
+      "step": 97500
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 0.08701920509338379,
+      "learning_rate": 0.00014314958357846704,
+      "loss": 3.422,
+      "step": 97600
+    },
+    {
+      "epoch": 0.977,
+      "grad_norm": 0.09414151310920715,
+      "learning_rate": 0.00014307630505634844,
+      "loss": 3.4204,
+      "step": 97700
+    },
+    {
+      "epoch": 0.978,
+      "grad_norm": 0.10199437290430069,
+      "learning_rate": 0.00014300313895335043,
+      "loss": 3.4202,
+      "step": 97800
+    },
+    {
+      "epoch": 0.979,
+      "grad_norm": 0.09039287269115448,
+      "learning_rate": 0.00014293008498232322,
+      "loss": 3.4193,
+      "step": 97900
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.10374122112989426,
+      "learning_rate": 0.00014285714285714284,
+      "loss": 3.4202,
+      "step": 98000
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 3.859375,
+      "eval_runtime": 32.984,
+      "eval_samples_per_second": 18256.958,
+      "eval_steps_per_second": 17.857,
+      "step": 98000
+    },
+    {
+      "epoch": 0.981,
+      "grad_norm": 0.0891217291355133,
+      "learning_rate": 0.00014278431229270645,
+      "loss": 3.4182,
+      "step": 98100
+    },
+    {
+      "epoch": 0.982,
+      "grad_norm": 0.0891917422413826,
+      "learning_rate": 0.0001427115930049275,
+      "loss": 3.4195,
+      "step": 98200
+    },
+    {
+      "epoch": 0.983,
+      "grad_norm": 0.11202235519886017,
+      "learning_rate": 0.0001426389847107313,
+      "loss": 3.4198,
+      "step": 98300
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 0.11552020907402039,
+      "learning_rate": 0.00014256648712805025,
+      "loss": 3.4204,
+      "step": 98400
+    },
+    {
+      "epoch": 0.985,
+      "grad_norm": 0.09355608373880386,
+      "learning_rate": 0.0001424940999758193,
+      "loss": 3.4167,
+      "step": 98500
+    },
+    {
+      "epoch": 0.986,
+      "grad_norm": 0.10058041661977768,
+      "learning_rate": 0.00014242182297397128,
+      "loss": 3.4191,
+      "step": 98600
+    },
+    {
+      "epoch": 0.987,
+      "grad_norm": 0.09189483523368835,
+      "learning_rate": 0.0001423496558434325,
+      "loss": 3.4181,
+      "step": 98700
+    },
+    {
+      "epoch": 0.988,
+      "grad_norm": 0.12277176976203918,
+      "learning_rate": 0.00014227759830611806,
+      "loss": 3.4168,
+      "step": 98800
+    },
+    {
+      "epoch": 0.989,
+      "grad_norm": 0.09363926202058792,
+      "learning_rate": 0.0001422056500849275,
+      "loss": 3.4179,
+      "step": 98900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.08723656833171844,
+      "learning_rate": 0.0001421338109037403,
+      "loss": 3.4195,
+      "step": 99000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 3.861328125,
+      "eval_runtime": 32.9644,
+      "eval_samples_per_second": 18267.85,
+      "eval_steps_per_second": 17.868,
+      "step": 99000
+    },
+    {
+      "epoch": 0.991,
+      "grad_norm": 0.08475708216428757,
+      "learning_rate": 0.00014206208048741123,
+      "loss": 3.4185,
+      "step": 99100
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 0.1032075509428978,
+      "learning_rate": 0.0001419904585617662,
+      "loss": 3.4171,
+      "step": 99200
+    },
+    {
+      "epoch": 0.993,
+      "grad_norm": 0.08446966111660004,
+      "learning_rate": 0.00014191894485359772,
+      "loss": 3.419,
+      "step": 99300
+    },
+    {
+      "epoch": 0.994,
+      "grad_norm": 0.08278857171535492,
+      "learning_rate": 0.0001418475390906605,
+      "loss": 3.4181,
+      "step": 99400
+    },
+    {
+      "epoch": 0.995,
+      "grad_norm": 0.10606228560209274,
+      "learning_rate": 0.00014177624100166717,
+      "loss": 3.4171,
+      "step": 99500
+    },
+    {
+      "epoch": 0.996,
+      "grad_norm": 0.09982995688915253,
+      "learning_rate": 0.00014170505031628393,
+      "loss": 3.4193,
+      "step": 99600
+    },
+    {
+      "epoch": 0.997,
+      "grad_norm": 0.08789572864770889,
+      "learning_rate": 0.0001416339667651262,
+      "loss": 3.4177,
+      "step": 99700
+    },
+    {
+      "epoch": 0.998,
+      "grad_norm": 0.08526172488927841,
+      "learning_rate": 0.0001415629900797544,
+      "loss": 3.4187,
+      "step": 99800
+    },
+    {
+      "epoch": 0.999,
+      "grad_norm": 0.13719169795513153,
+      "learning_rate": 0.00014149211999266962,
+      "loss": 3.4189,
+      "step": 99900
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.08984668552875519,
+      "learning_rate": 0.0001414213562373095,
+      "loss": 3.4164,
+      "step": 100000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 3.859375,
+      "eval_runtime": 32.9957,
+      "eval_samples_per_second": 18250.5,
+      "eval_steps_per_second": 17.851,
+      "step": 100000
+    },
+    {
+      "epoch": 1.0,
+      "step": 100000,
+      "total_flos": 8.175126367943472e+19,
+      "train_loss": 3.5368977130126953,
+      "train_runtime": 127327.336,
+      "train_samples_per_second": 3920.604,
+      "train_steps_per_second": 0.785
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.175126367943472e+19,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}