diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5669 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.944375772558715,
+  "eval_steps": 5000,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00024721878862793575,
+      "grad_norm": 1448.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 13.9716,
+      "step": 1
+    },
+    {
+      "epoch": 0.006180469715698393,
+      "grad_norm": 334.0,
+      "learning_rate": 5e-06,
+      "loss": 9.5244,
+      "step": 25
+    },
+    {
+      "epoch": 0.012360939431396786,
+      "grad_norm": 78.5,
+      "learning_rate": 1e-05,
+      "loss": 7.5535,
+      "step": 50
+    },
+    {
+      "epoch": 0.018541409147095178,
+      "grad_norm": 34.5,
+      "learning_rate": 1.5e-05,
+      "loss": 7.0022,
+      "step": 75
+    },
+    {
+      "epoch": 0.024721878862793572,
+      "grad_norm": 18.75,
+      "learning_rate": 2e-05,
+      "loss": 6.4562,
+      "step": 100
+    },
+    {
+      "epoch": 0.030902348578491966,
+      "grad_norm": 63.75,
+      "learning_rate": 2.5e-05,
+      "loss": 5.6999,
+      "step": 125
+    },
+    {
+      "epoch": 0.037082818294190356,
+      "grad_norm": 29.125,
+      "learning_rate": 3e-05,
+      "loss": 5.2536,
+      "step": 150
+    },
+    {
+      "epoch": 0.04326328800988875,
+      "grad_norm": 29.5,
+      "learning_rate": 3.5e-05,
+      "loss": 4.6767,
+      "step": 175
+    },
+    {
+      "epoch": 0.049443757725587144,
+      "grad_norm": 20.25,
+      "learning_rate": 4e-05,
+      "loss": 4.2913,
+      "step": 200
+    },
+    {
+      "epoch": 0.05562422744128554,
+      "grad_norm": 16.625,
+      "learning_rate": 4.5e-05,
+      "loss": 3.7505,
+      "step": 225
+    },
+    {
+      "epoch": 0.06180469715698393,
+      "grad_norm": 17.375,
+      "learning_rate": 5e-05,
+      "loss": 3.3317,
+      "step": 250
+    },
+    {
+      "epoch": 0.06798516687268233,
+      "grad_norm": 112.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.1318,
+      "step": 275
+    },
+    {
+      "epoch": 0.07416563658838071,
+      "grad_norm": 31.75,
+      "learning_rate": 6e-05,
+      "loss": 3.558,
+      "step": 300
+    },
+    {
+      "epoch": 0.08034610630407911,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 2.9825,
+      "step": 325
+    },
+    {
+      "epoch": 0.0865265760197775,
+      "grad_norm": 12.875,
+      "learning_rate": 7e-05,
+      "loss": 2.6931,
+      "step": 350
+    },
+    {
+      "epoch": 0.09270704573547589,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.4328,
+      "step": 375
+    },
+    {
+      "epoch": 0.09888751545117429,
+      "grad_norm": 11.0625,
+      "learning_rate": 8e-05,
+      "loss": 2.285,
+      "step": 400
+    },
+    {
+      "epoch": 0.10506798516687268,
+      "grad_norm": 6.28125,
+      "learning_rate": 8.5e-05,
+      "loss": 2.1667,
+      "step": 425
+    },
+    {
+      "epoch": 0.11124845488257108,
+      "grad_norm": 4.5625,
+      "learning_rate": 9e-05,
+      "loss": 2.0534,
+      "step": 450
+    },
+    {
+      "epoch": 0.11742892459826947,
+      "grad_norm": 3.90625,
+      "learning_rate": 9.5e-05,
+      "loss": 1.9744,
+      "step": 475
+    },
+    {
+      "epoch": 0.12360939431396786,
+      "grad_norm": 4.78125,
+      "learning_rate": 0.0001,
+      "loss": 1.9209,
+      "step": 500
+    },
+    {
+      "epoch": 0.12978986402966625,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.987179487179488e-05,
+      "loss": 1.8793,
+      "step": 525
+    },
+    {
+      "epoch": 0.13597033374536466,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.974358974358975e-05,
+      "loss": 1.8301,
+      "step": 550
+    },
+    {
+      "epoch": 0.14215080346106304,
+      "grad_norm": 2.25,
+      "learning_rate": 9.961538461538463e-05,
+      "loss": 1.7935,
+      "step": 575
+    },
+    {
+      "epoch": 0.14833127317676142,
+      "grad_norm": 2.3125,
+      "learning_rate": 9.948717948717949e-05,
+      "loss": 1.7651,
+      "step": 600
+    },
+    {
+      "epoch": 0.15451174289245984,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.935897435897437e-05,
+      "loss": 1.737,
+      "step": 625
+    },
+    {
+      "epoch": 0.16069221260815822,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.923076923076923e-05,
+      "loss": 1.7098,
+      "step": 650
+    },
+    {
+      "epoch": 0.1668726823238566,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.910256410256411e-05,
+      "loss": 1.69,
+      "step": 675
+    },
+    {
+      "epoch": 0.173053152039555,
+      "grad_norm": 1.671875,
+      "learning_rate": 9.897435897435898e-05,
+      "loss": 1.6622,
+      "step": 700
+    },
+    {
+      "epoch": 0.1792336217552534,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.884615384615386e-05,
+      "loss": 1.6502,
+      "step": 725
+    },
+    {
+      "epoch": 0.18541409147095178,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 1.629,
+      "step": 750
+    },
+    {
+      "epoch": 0.1915945611866502,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.85897435897436e-05,
+      "loss": 1.6102,
+      "step": 775
+    },
+    {
+      "epoch": 0.19777503090234858,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.846153846153848e-05,
+      "loss": 1.5991,
+      "step": 800
+    },
+    {
+      "epoch": 0.20395550061804696,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.833333333333333e-05,
+      "loss": 1.5884,
+      "step": 825
+    },
+    {
+      "epoch": 0.21013597033374537,
+      "grad_norm": 1.5,
+      "learning_rate": 9.820512820512821e-05,
+      "loss": 1.5704,
+      "step": 850
+    },
+    {
+      "epoch": 0.21631644004944375,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.807692307692307e-05,
+      "loss": 1.5573,
+      "step": 875
+    },
+    {
+      "epoch": 0.22249690976514216,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.794871794871795e-05,
+      "loss": 1.5416,
+      "step": 900
+    },
+    {
+      "epoch": 0.22867737948084055,
+      "grad_norm": 1.25,
+      "learning_rate": 9.782051282051282e-05,
+      "loss": 1.5409,
+      "step": 925
+    },
+    {
+      "epoch": 0.23485784919653893,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.76923076923077e-05,
+      "loss": 1.5344,
+      "step": 950
+    },
+    {
+      "epoch": 0.24103831891223734,
+      "grad_norm": 1.5,
+      "learning_rate": 9.756410256410257e-05,
+      "loss": 1.5195,
+      "step": 975
+    },
+    {
+      "epoch": 0.24721878862793573,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 1.5051,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25339925834363414,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.730769230769232e-05,
+      "loss": 1.505,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2595797280593325,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.717948717948718e-05,
+      "loss": 1.4932,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2657601977750309,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.705128205128206e-05,
+      "loss": 1.4901,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2719406674907293,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.692307692307692e-05,
+      "loss": 1.4806,
+      "step": 1100
+    },
+    {
+      "epoch": 0.27812113720642767,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.67948717948718e-05,
+      "loss": 1.4745,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2843016069221261,
+      "grad_norm": 1.3125,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 1.4604,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2904820766378245,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.653846153846155e-05,
+      "loss": 1.4584,
+      "step": 1175
+    },
+    {
+      "epoch": 0.29666254635352285,
+      "grad_norm": 1.5234375,
+      "learning_rate": 9.641025641025641e-05,
+      "loss": 1.4489,
+      "step": 1200
+    },
+    {
+      "epoch": 0.30284301606922126,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.628205128205129e-05,
+      "loss": 1.4439,
+      "step": 1225
+    },
+    {
+      "epoch": 0.30902348578491967,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 1.4367,
+      "step": 1250
+    },
+    {
+      "epoch": 0.315203955500618,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.602564102564103e-05,
+      "loss": 1.4392,
+      "step": 1275
+    },
+    {
+      "epoch": 0.32138442521631644,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.589743589743591e-05,
+      "loss": 1.4281,
+      "step": 1300
+    },
+    {
+      "epoch": 0.32756489493201485,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.576923076923078e-05,
+      "loss": 1.4251,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3337453646477132,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.564102564102565e-05,
+      "loss": 1.4222,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3399258343634116,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.551282051282052e-05,
+      "loss": 1.4259,
+      "step": 1375
+    },
+    {
+      "epoch": 0.34610630407911,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.53846153846154e-05,
+      "loss": 1.4154,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3522867737948084,
+      "grad_norm": 1.0,
+      "learning_rate": 9.525641025641026e-05,
+      "loss": 1.4074,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3584672435105068,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.512820512820513e-05,
+      "loss": 1.4051,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3646477132262052,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.5e-05,
+      "loss": 1.4007,
+      "step": 1475
+    },
+    {
+      "epoch": 0.37082818294190356,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 1.389,
+      "step": 1500
+    },
+    {
+      "epoch": 0.377008652657602,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.474358974358975e-05,
+      "loss": 1.3924,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3831891223733004,
+      "grad_norm": 0.890625,
+      "learning_rate": 9.461538461538461e-05,
+      "loss": 1.3868,
+      "step": 1550
+    },
+    {
+      "epoch": 0.38936959208899874,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.448717948717949e-05,
+      "loss": 1.3889,
+      "step": 1575
+    },
+    {
+      "epoch": 0.39555006180469715,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.435897435897436e-05,
+      "loss": 1.3825,
+      "step": 1600
+    },
+    {
+      "epoch": 0.40173053152039556,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.423076923076924e-05,
+      "loss": 1.3709,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4079110012360939,
+      "grad_norm": 0.9140625,
+      "learning_rate": 9.41025641025641e-05,
+      "loss": 1.3719,
+      "step": 1650
+    },
+    {
+      "epoch": 0.41409147095179233,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.397435897435898e-05,
+      "loss": 1.3718,
+      "step": 1675
+    },
+    {
+      "epoch": 0.42027194066749074,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.384615384615386e-05,
+      "loss": 1.3673,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4264524103831891,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.371794871794872e-05,
+      "loss": 1.3629,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4326328800988875,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 1.3616,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4388133498145859,
+      "grad_norm": 0.89453125,
+      "learning_rate": 9.346153846153846e-05,
+      "loss": 1.3599,
+      "step": 1775
+    },
+    {
+      "epoch": 0.44499381953028433,
+      "grad_norm": 0.88671875,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 1.3516,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4511742892459827,
+      "grad_norm": 0.796875,
+      "learning_rate": 9.320512820512821e-05,
+      "loss": 1.3552,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4573547589616811,
+      "grad_norm": 0.89453125,
+      "learning_rate": 9.307692307692309e-05,
+      "loss": 1.3448,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4635352286773795,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.294871794871795e-05,
+      "loss": 1.3492,
+      "step": 1875
+    },
+    {
+      "epoch": 0.46971569839307786,
+      "grad_norm": 0.85546875,
+      "learning_rate": 9.282051282051283e-05,
+      "loss": 1.34,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4758961681087763,
+      "grad_norm": 0.7734375,
+      "learning_rate": 9.26923076923077e-05,
+      "loss": 1.3406,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4820766378244747,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.256410256410257e-05,
+      "loss": 1.3369,
+      "step": 1950
+    },
+    {
+      "epoch": 0.48825710754017304,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.243589743589745e-05,
+      "loss": 1.3297,
+      "step": 1975
+    },
+    {
+      "epoch": 0.49443757725587145,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 1.3347,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5006180469715699,
+      "grad_norm": 0.81640625,
+      "learning_rate": 9.217948717948718e-05,
+      "loss": 1.3269,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5067985166872683,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.205128205128205e-05,
+      "loss": 1.3316,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5129789864029666,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.192307692307692e-05,
+      "loss": 1.3212,
+      "step": 2075
+    },
+    {
+      "epoch": 0.519159456118665,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.179487179487179e-05,
+      "loss": 1.3177,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5253399258343634,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 1.3155,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5315203955500618,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.153846153846155e-05,
+      "loss": 1.3206,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5377008652657602,
+      "grad_norm": 0.796875,
+      "learning_rate": 9.141025641025641e-05,
+      "loss": 1.3131,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5438813349814586,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.128205128205129e-05,
+      "loss": 1.3143,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5500618046971569,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.115384615384615e-05,
+      "loss": 1.3099,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5562422744128553,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 1.3104,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5624227441285538,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.08974358974359e-05,
+      "loss": 1.3026,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5686032138442522,
+      "grad_norm": 0.75390625,
+      "learning_rate": 9.076923076923078e-05,
+      "loss": 1.2985,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5747836835599506,
+      "grad_norm": 0.80078125,
+      "learning_rate": 9.064102564102564e-05,
+      "loss": 1.3109,
+      "step": 2325
+    },
+    {
+      "epoch": 0.580964153275649,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.051282051282052e-05,
+      "loss": 1.3061,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5871446229913473,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.038461538461538e-05,
+      "loss": 1.3022,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5933250927070457,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.025641025641026e-05,
+      "loss": 1.3006,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5995055624227441,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.012820512820514e-05,
+      "loss": 1.2938,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6056860321384425,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9e-05,
+      "loss": 1.2968,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6118665018541409,
+      "grad_norm": 0.8046875,
+      "learning_rate": 8.987179487179488e-05,
+      "loss": 1.2872,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6180469715698393,
+      "grad_norm": 0.828125,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 1.2902,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6242274412855378,
+      "grad_norm": 0.88671875,
+      "learning_rate": 8.961538461538463e-05,
+      "loss": 1.2848,
+      "step": 2525
+    },
+    {
+      "epoch": 0.630407911001236,
+      "grad_norm": 0.76171875,
+      "learning_rate": 8.948717948717949e-05,
+      "loss": 1.2902,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6365883807169345,
+      "grad_norm": 0.859375,
+      "learning_rate": 8.935897435897437e-05,
+      "loss": 1.2917,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6427688504326329,
+      "grad_norm": 0.76953125,
+      "learning_rate": 8.923076923076924e-05,
+      "loss": 1.2882,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6489493201483313,
+      "grad_norm": 0.84375,
+      "learning_rate": 8.910256410256411e-05,
+      "loss": 1.277,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6551297898640297,
+      "grad_norm": 0.71875,
+      "learning_rate": 8.897435897435898e-05,
+      "loss": 1.2832,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6613102595797281,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.884615384615384e-05,
+      "loss": 1.2822,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6674907292954264,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.871794871794872e-05,
+      "loss": 1.2809,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6736711990111248,
+      "grad_norm": 0.859375,
+      "learning_rate": 8.858974358974359e-05,
+      "loss": 1.2696,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6798516687268232,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 1.2768,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6860321384425216,
+      "grad_norm": 0.8515625,
+      "learning_rate": 8.833333333333333e-05,
+      "loss": 1.2735,
+      "step": 2775
+    },
+    {
+      "epoch": 0.69221260815822,
+      "grad_norm": 0.7734375,
+      "learning_rate": 8.820512820512821e-05,
+      "loss": 1.2743,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6983930778739185,
+      "grad_norm": 0.796875,
+      "learning_rate": 8.807692307692307e-05,
+      "loss": 1.2766,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7045735475896168,
+      "grad_norm": 0.81640625,
+      "learning_rate": 8.794871794871795e-05,
+      "loss": 1.2693,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7107540173053152,
+      "grad_norm": 0.83203125,
+      "learning_rate": 8.782051282051283e-05,
+      "loss": 1.263,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7169344870210136,
+      "grad_norm": 0.80859375,
+      "learning_rate": 8.76923076923077e-05,
+      "loss": 1.2618,
+      "step": 2900
+    },
+    {
+      "epoch": 0.723114956736712,
+      "grad_norm": 0.77734375,
+      "learning_rate": 8.756410256410257e-05,
+      "loss": 1.2626,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7292954264524104,
+      "grad_norm": 0.8203125,
+      "learning_rate": 8.743589743589744e-05,
+      "loss": 1.2644,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7354758961681088,
+      "grad_norm": 0.7734375,
+      "learning_rate": 8.730769230769232e-05,
+      "loss": 1.2623,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7416563658838071,
+      "grad_norm": 0.77734375,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 1.2633,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7478368355995055,
+      "grad_norm": 0.75,
+      "learning_rate": 8.705128205128206e-05,
+      "loss": 1.261,
+      "step": 3025
+    },
+    {
+      "epoch": 0.754017305315204,
+      "grad_norm": 0.7890625,
+      "learning_rate": 8.692307692307692e-05,
+      "loss": 1.259,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7601977750309024,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.67948717948718e-05,
+      "loss": 1.2546,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7663782447466008,
+      "grad_norm": 0.71875,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 1.2551,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7725587144622992,
+      "grad_norm": 0.71484375,
+      "learning_rate": 8.653846153846155e-05,
+      "loss": 1.2565,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7787391841779975,
+      "grad_norm": 0.76953125,
+      "learning_rate": 8.641025641025642e-05,
+      "loss": 1.2559,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7849196538936959,
+      "grad_norm": 0.76953125,
+      "learning_rate": 8.628205128205129e-05,
+      "loss": 1.253,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7911001236093943,
+      "grad_norm": 0.83203125,
+      "learning_rate": 8.615384615384617e-05,
+      "loss": 1.255,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7972805933250927,
+      "grad_norm": 0.72265625,
+      "learning_rate": 8.602564102564103e-05,
+      "loss": 1.2486,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8034610630407911,
+      "grad_norm": 0.74609375,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 1.24,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8096415327564895,
+      "grad_norm": 0.8125,
+      "learning_rate": 8.576923076923076e-05,
+      "loss": 1.2466,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8158220024721878,
+      "grad_norm": 0.765625,
+      "learning_rate": 8.564102564102564e-05,
+      "loss": 1.2455,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8220024721878862,
+      "grad_norm": 0.76171875,
+      "learning_rate": 8.551282051282052e-05,
+      "loss": 1.2409,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8281829419035847,
+      "grad_norm": 0.875,
+      "learning_rate": 8.538461538461538e-05,
+      "loss": 1.237,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8343634116192831,
+      "grad_norm": 0.82421875,
+      "learning_rate": 8.525641025641026e-05,
+      "loss": 1.2437,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8405438813349815,
+      "grad_norm": 0.73828125,
+      "learning_rate": 8.512820512820513e-05,
+      "loss": 1.2388,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8467243510506799,
+      "grad_norm": 0.8671875,
+      "learning_rate": 8.5e-05,
+      "loss": 1.238,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8529048207663782,
+      "grad_norm": 0.76171875,
+      "learning_rate": 8.487179487179487e-05,
+      "loss": 1.231,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8590852904820766,
+      "grad_norm": 0.7421875,
+      "learning_rate": 8.474358974358975e-05,
+      "loss": 1.2381,
+      "step": 3475
+    },
+    {
+      "epoch": 0.865265760197775,
+      "grad_norm": 0.75390625,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 1.2305,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8714462299134734,
+      "grad_norm": 0.70703125,
+      "learning_rate": 8.448717948717949e-05,
+      "loss": 1.2326,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8776266996291718,
+      "grad_norm": 0.8046875,
+      "learning_rate": 8.435897435897436e-05,
+      "loss": 1.2288,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8838071693448702,
+      "grad_norm": 0.828125,
+      "learning_rate": 8.423076923076924e-05,
+      "loss": 1.2268,
+      "step": 3575
+    },
+    {
+      "epoch": 0.8899876390605687,
+      "grad_norm": 0.7421875,
+      "learning_rate": 8.410256410256411e-05,
+      "loss": 1.2314,
+      "step": 3600
+    },
+    {
+      "epoch": 0.896168108776267,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.397435897435898e-05,
+      "loss": 1.2292,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9023485784919654,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.384615384615386e-05,
+      "loss": 1.2304,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9085290482076638,
+      "grad_norm": 0.765625,
+      "learning_rate": 8.371794871794872e-05,
+      "loss": 1.2256,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9147095179233622,
+      "grad_norm": 0.765625,
+      "learning_rate": 8.35897435897436e-05,
+      "loss": 1.2268,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9208899876390606,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.346153846153847e-05,
+      "loss": 1.2287,
+      "step": 3725
+    },
+    {
+      "epoch": 0.927070457354759,
+      "grad_norm": 0.72265625,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.2252,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9332509270704573,
+      "grad_norm": 0.78515625,
+      "learning_rate": 8.320512820512821e-05,
+      "loss": 1.225,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9394313967861557,
+      "grad_norm": 0.7734375,
+      "learning_rate": 8.307692307692309e-05,
+      "loss": 1.2183,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9456118665018541,
+      "grad_norm": 0.6796875,
+      "learning_rate": 8.294871794871795e-05,
+      "loss": 1.2244,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9517923362175525,
+      "grad_norm": 0.71484375,
+      "learning_rate": 8.282051282051283e-05,
+      "loss": 1.2194,
+      "step": 3850
+    },
+    {
+      "epoch": 0.957972805933251,
+      "grad_norm": 0.75,
+      "learning_rate": 8.26923076923077e-05,
+      "loss": 1.2162,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9641532756489494,
+      "grad_norm": 0.73828125,
+      "learning_rate": 8.256410256410256e-05,
+      "loss": 1.2146,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9703337453646477,
+      "grad_norm": 0.76953125,
+      "learning_rate": 8.243589743589744e-05,
+      "loss": 1.2144,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9765142150803461,
+      "grad_norm": 0.73828125,
+      "learning_rate": 8.23076923076923e-05,
+      "loss": 1.2146,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9826946847960445,
+      "grad_norm": 0.70703125,
+      "learning_rate": 8.217948717948718e-05,
+      "loss": 1.2151,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9888751545117429,
+      "grad_norm": 0.72265625,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 1.2146,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9950556242274413,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.192307692307693e-05,
+      "loss": 1.211,
+      "step": 4025
+    },
+    {
+      "epoch": 1.0012360939431397,
+      "grad_norm": 0.88671875,
+      "learning_rate": 8.179487179487179e-05,
+      "loss": 1.198,
+      "step": 4050
+    },
+    {
+      "epoch": 1.0074165636588381,
+      "grad_norm": 0.8046875,
+      "learning_rate": 8.166666666666667e-05,
+      "loss": 1.1373,
+      "step": 4075
+    },
+    {
+      "epoch": 1.0135970333745365,
+      "grad_norm": 0.84375,
+      "learning_rate": 8.153846153846155e-05,
+      "loss": 1.1436,
+      "step": 4100
+    },
+    {
+      "epoch": 1.019777503090235,
+      "grad_norm": 0.7265625,
+      "learning_rate": 8.141025641025641e-05,
+      "loss": 1.138,
+      "step": 4125
+    },
+    {
+      "epoch": 1.0259579728059331,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.128205128205129e-05,
+      "loss": 1.1334,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0321384425216316,
+      "grad_norm": 0.828125,
+      "learning_rate": 8.115384615384616e-05,
+      "loss": 1.1423,
+      "step": 4175
+    },
+    {
+      "epoch": 1.03831891223733,
+      "grad_norm": 0.75390625,
+      "learning_rate": 8.102564102564103e-05,
+      "loss": 1.1398,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0444993819530284,
+      "grad_norm": 0.7890625,
+      "learning_rate": 8.08974358974359e-05,
+      "loss": 1.1378,
+      "step": 4225
+    },
+    {
+      "epoch": 1.0506798516687268,
+      "grad_norm": 0.76953125,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 1.1376,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0568603213844252,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.064102564102564e-05,
+      "loss": 1.1443,
+      "step": 4275
+    },
+    {
+      "epoch": 1.0630407911001236,
+      "grad_norm": 0.828125,
+      "learning_rate": 8.051282051282052e-05,
+      "loss": 1.1441,
+      "step": 4300
+    },
+    {
+      "epoch": 1.069221260815822,
+      "grad_norm": 0.77734375,
+      "learning_rate": 8.038461538461538e-05,
+      "loss": 1.1435,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0754017305315204,
+      "grad_norm": 0.73828125,
+      "learning_rate": 8.025641025641026e-05,
+      "loss": 1.1449,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0815822002472189,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.012820512820514e-05,
+      "loss": 1.1468,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0877626699629173,
+      "grad_norm": 0.796875,
+      "learning_rate": 8e-05,
+      "loss": 1.1402,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0939431396786157,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.987179487179488e-05,
+      "loss": 1.1405,
+      "step": 4425
+    },
+    {
+      "epoch": 1.100123609394314,
+      "grad_norm": 0.7421875,
+      "learning_rate": 7.974358974358975e-05,
+      "loss": 1.1441,
+      "step": 4450
+    },
+    {
+      "epoch": 1.1063040791100123,
+      "grad_norm": 0.74609375,
+      "learning_rate": 7.961538461538461e-05,
+      "loss": 1.1486,
+      "step": 4475
+    },
+    {
+      "epoch": 1.1124845488257107,
+      "grad_norm": 0.7890625,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 1.1439,
+      "step": 4500
+    },
+    {
+      "epoch": 1.118665018541409,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.935897435897436e-05,
+      "loss": 1.1408,
+      "step": 4525
+    },
+    {
+      "epoch": 1.1248454882571075,
+      "grad_norm": 0.74609375,
+      "learning_rate": 7.923076923076924e-05,
+      "loss": 1.1381,
+      "step": 4550
+    },
+    {
+      "epoch": 1.131025957972806,
+      "grad_norm": 0.77734375,
+      "learning_rate": 7.91025641025641e-05,
+      "loss": 1.1441,
+      "step": 4575
+    },
+    {
+      "epoch": 1.1372064276885043,
+      "grad_norm": 0.80078125,
+      "learning_rate": 7.897435897435898e-05,
+      "loss": 1.1377,
+      "step": 4600
+    },
+    {
+      "epoch": 1.1433868974042027,
+      "grad_norm": 0.7734375,
+      "learning_rate": 7.884615384615384e-05,
+      "loss": 1.1405,
+      "step": 4625
+    },
+    {
+      "epoch": 1.1495673671199012,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.871794871794872e-05,
+      "loss": 1.146,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1557478368355996,
+      "grad_norm": 0.8125,
+      "learning_rate": 7.858974358974359e-05,
+      "loss": 1.1416,
+      "step": 4675
+    },
+    {
+      "epoch": 1.161928306551298,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.846153846153847e-05,
+      "loss": 1.1352,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1681087762669964,
+      "grad_norm": 0.83984375,
+      "learning_rate": 7.833333333333333e-05,
+      "loss": 1.1398,
+      "step": 4725
+    },
+    {
+      "epoch": 1.1742892459826946,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 1.1422,
+      "step": 4750
+    },
+    {
+      "epoch": 1.180469715698393,
+      "grad_norm": 0.72265625,
+      "learning_rate": 7.807692307692307e-05,
+      "loss": 1.1416,
+      "step": 4775
+    },
+    {
+      "epoch": 1.1866501854140914,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.794871794871795e-05,
+      "loss": 1.1416,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1928306551297898,
+      "grad_norm": 0.80078125,
+      "learning_rate": 7.782051282051283e-05,
+      "loss": 1.1387,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1990111248454882,
+      "grad_norm": 0.7890625,
+      "learning_rate": 7.76923076923077e-05,
+      "loss": 1.1372,
+      "step": 4850
+    },
+    {
+      "epoch": 1.2051915945611866,
+      "grad_norm": 0.75,
+      "learning_rate": 7.756410256410257e-05,
+      "loss": 1.1381,
+      "step": 4875
+    },
+    {
+      "epoch": 1.211372064276885,
+      "grad_norm": 0.72265625,
+      "learning_rate": 7.743589743589744e-05,
+      "loss": 1.1342,
+      "step": 4900
+    },
+    {
+      "epoch": 1.2175525339925835,
+      "grad_norm": 0.71875,
+      "learning_rate": 7.730769230769232e-05,
+      "loss": 1.1355,
+      "step": 4925
+    },
+    {
+      "epoch": 1.2237330037082819,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.717948717948718e-05,
+      "loss": 1.1406,
+      "step": 4950
+    },
+    {
+      "epoch": 1.2299134734239803,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.705128205128206e-05,
+      "loss": 1.1411,
+      "step": 4975
+    },
+    {
+      "epoch": 1.2360939431396787,
+      "grad_norm": 0.7578125,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 1.135,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2360939431396787,
+      "eval_loss": 1.048352837562561,
+      "eval_runtime": 1.5386,
+      "eval_samples_per_second": 415.323,
+      "eval_steps_per_second": 1.95,
+      "step": 5000
+    },
+    {
+      "epoch": 1.242274412855377,
+      "grad_norm": 0.8203125,
+      "learning_rate": 7.67948717948718e-05,
+      "loss": 1.1348,
+      "step": 5025
+    },
+    {
+      "epoch": 1.2484548825710755,
+      "grad_norm": 0.7578125,
+      "learning_rate": 7.666666666666667e-05,
+      "loss": 1.1338,
+      "step": 5050
+    },
+    {
+      "epoch": 1.254635352286774,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.653846153846153e-05,
+      "loss": 1.1291,
+      "step": 5075
+    },
+    {
+      "epoch": 1.260815822002472,
+      "grad_norm": 0.7734375,
+      "learning_rate": 7.641025641025641e-05,
+      "loss": 1.1325,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2669962917181705,
+      "grad_norm": 0.7265625,
+      "learning_rate": 7.628205128205128e-05,
+      "loss": 1.1359,
+      "step": 5125
+    },
+    {
+      "epoch": 1.273176761433869,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.615384615384616e-05,
+      "loss": 1.1337,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2793572311495673,
+      "grad_norm": 0.7890625,
+      "learning_rate": 7.602564102564102e-05,
+      "loss": 1.132,
+      "step": 5175
+    },
+    {
+      "epoch": 1.2855377008652658,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.58974358974359e-05,
+      "loss": 1.1308,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2917181705809642,
+      "grad_norm": 0.75,
+      "learning_rate": 7.576923076923076e-05,
+      "loss": 1.1291,
+      "step": 5225
+    },
+    {
+      "epoch": 1.2978986402966626,
+      "grad_norm": 0.7109375,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 1.1248,
+      "step": 5250
+    },
+    {
+      "epoch": 1.304079110012361,
+      "grad_norm": 0.78515625,
+      "learning_rate": 7.551282051282052e-05,
+      "loss": 1.1262,
+      "step": 5275
+    },
+    {
+      "epoch": 1.3102595797280594,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.538461538461539e-05,
+      "loss": 1.1246,
+      "step": 5300
+    },
+    {
+      "epoch": 1.3164400494437576,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.525641025641026e-05,
+      "loss": 1.1234,
+      "step": 5325
+    },
+    {
+      "epoch": 1.322620519159456,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.512820512820513e-05,
+      "loss": 1.1252,
+      "step": 5350
+    },
+    {
+      "epoch": 1.3288009888751544,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.1288,
+      "step": 5375
+    },
+    {
+      "epoch": 1.3349814585908528,
+      "grad_norm": 0.7265625,
+      "learning_rate": 7.487179487179487e-05,
+      "loss": 1.1227,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3411619283065512,
+      "grad_norm": 0.72265625,
+      "learning_rate": 7.474358974358975e-05,
+      "loss": 1.1233,
+      "step": 5425
+    },
+    {
+      "epoch": 1.3473423980222496,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.461538461538462e-05,
+      "loss": 1.1233,
+      "step": 5450
+    },
+    {
+      "epoch": 1.353522867737948,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.44871794871795e-05,
+      "loss": 1.1248,
+      "step": 5475
+    },
+    {
+      "epoch": 1.3597033374536465,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 1.1264,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3658838071693449,
+      "grad_norm": 0.703125,
+      "learning_rate": 7.423076923076924e-05,
+      "loss": 1.1215,
+      "step": 5525
+    },
+    {
+      "epoch": 1.3720642768850433,
+      "grad_norm": 0.69921875,
+      "learning_rate": 7.410256410256412e-05,
+      "loss": 1.1237,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3782447466007417,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.397435897435898e-05,
+      "loss": 1.1212,
+      "step": 5575
+    },
+    {
+      "epoch": 1.38442521631644,
+      "grad_norm": 0.7578125,
+      "learning_rate": 7.384615384615386e-05,
+      "loss": 1.118,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3906056860321385,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.371794871794872e-05,
+      "loss": 1.1163,
+      "step": 5625
+    },
+    {
+      "epoch": 1.396786155747837,
+      "grad_norm": 0.73046875,
+      "learning_rate": 7.35897435897436e-05,
+      "loss": 1.1168,
+      "step": 5650
+    },
+    {
+      "epoch": 1.4029666254635353,
+      "grad_norm": 0.66796875,
+      "learning_rate": 7.346153846153847e-05,
+      "loss": 1.118,
+      "step": 5675
+    },
+    {
+      "epoch": 1.4091470951792338,
+      "grad_norm": 0.79296875,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 1.1153,
+      "step": 5700
+    },
+    {
+      "epoch": 1.415327564894932,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.320512820512821e-05,
+      "loss": 1.1237,
+      "step": 5725
+    },
+    {
+      "epoch": 1.4215080346106304,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 1.1225,
+      "step": 5750
+    },
+    {
+      "epoch": 1.4276885043263288,
+      "grad_norm": 0.765625,
+      "learning_rate": 7.294871794871795e-05,
+      "loss": 1.1224,
+      "step": 5775
+    },
+    {
+      "epoch": 1.4338689740420272,
+      "grad_norm": 0.76171875,
+      "learning_rate": 7.282051282051282e-05,
+      "loss": 1.1187,
+      "step": 5800
+    },
+    {
+      "epoch": 1.4400494437577256,
+      "grad_norm": 0.703125,
+      "learning_rate": 7.26923076923077e-05,
+      "loss": 1.1186,
+      "step": 5825
+    },
+    {
+      "epoch": 1.446229913473424,
+      "grad_norm": 0.79296875,
+      "learning_rate": 7.256410256410256e-05,
+      "loss": 1.1164,
+      "step": 5850
+    },
+    {
+      "epoch": 1.4524103831891224,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.243589743589744e-05,
+      "loss": 1.1194,
+      "step": 5875
+    },
+    {
+      "epoch": 1.4585908529048208,
+      "grad_norm": 0.7734375,
+      "learning_rate": 7.23076923076923e-05,
+      "loss": 1.1164,
+      "step": 5900
+    },
+    {
+      "epoch": 1.4647713226205192,
+      "grad_norm": 0.7109375,
+      "learning_rate": 7.217948717948718e-05,
+      "loss": 1.1159,
+      "step": 5925
+    },
+    {
+      "epoch": 1.4709517923362174,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.205128205128205e-05,
+      "loss": 1.1184,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4771322620519158,
+      "grad_norm": 0.734375,
+      "learning_rate": 7.192307692307693e-05,
+      "loss": 1.1136,
+      "step": 5975
+    },
+    {
+      "epoch": 1.4833127317676142,
+      "grad_norm": 0.77734375,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 1.1086,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4894932014833127,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.166666666666667e-05,
+      "loss": 1.1098,
+      "step": 6025
+    },
+    {
+      "epoch": 1.495673671199011,
+      "grad_norm": 0.8046875,
+      "learning_rate": 7.153846153846155e-05,
+      "loss": 1.1132,
+      "step": 6050
+    },
+    {
+      "epoch": 1.5018541409147095,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.141025641025641e-05,
+      "loss": 1.1044,
+      "step": 6075
+    },
+    {
+      "epoch": 1.508034610630408,
+      "grad_norm": 0.74609375,
+      "learning_rate": 7.128205128205129e-05,
+      "loss": 1.1081,
+      "step": 6100
+    },
+    {
+      "epoch": 1.5142150803461063,
+      "grad_norm": 0.74609375,
+      "learning_rate": 7.115384615384616e-05,
+      "loss": 1.1109,
+      "step": 6125
+    },
+    {
+      "epoch": 1.5203955500618047,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.102564102564103e-05,
+      "loss": 1.1085,
+      "step": 6150
+    },
+    {
+      "epoch": 1.5265760197775031,
+      "grad_norm": 0.80859375,
+      "learning_rate": 7.08974358974359e-05,
+      "loss": 1.1143,
+      "step": 6175
+    },
+    {
+      "epoch": 1.5327564894932015,
+      "grad_norm": 0.73046875,
+      "learning_rate": 7.076923076923078e-05,
+      "loss": 1.106,
+      "step": 6200
+    },
+    {
+      "epoch": 1.5389369592089,
+      "grad_norm": 0.7421875,
+      "learning_rate": 7.064102564102564e-05,
+      "loss": 1.1094,
+      "step": 6225
+    },
+    {
+      "epoch": 1.5451174289245984,
+      "grad_norm": 0.8203125,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 1.1063,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5512978986402968,
+      "grad_norm": 0.7265625,
+      "learning_rate": 7.03846153846154e-05,
+      "loss": 1.105,
+      "step": 6275
+    },
+    {
+      "epoch": 1.5574783683559952,
+      "grad_norm": 0.7421875,
+      "learning_rate": 7.025641025641025e-05,
+      "loss": 1.1063,
+      "step": 6300
+    },
+    {
+      "epoch": 1.5636588380716936,
+      "grad_norm": 0.7421875,
+      "learning_rate": 7.012820512820513e-05,
+      "loss": 1.1029,
+      "step": 6325
+    },
+    {
+      "epoch": 1.569839307787392,
+      "grad_norm": 0.78515625,
+      "learning_rate": 7e-05,
+      "loss": 1.105,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5760197775030902,
+      "grad_norm": 0.75,
+      "learning_rate": 6.987179487179487e-05,
+      "loss": 1.1036,
+      "step": 6375
+    },
+    {
+      "epoch": 1.5822002472187886,
+      "grad_norm": 0.73828125,
+      "learning_rate": 6.974358974358974e-05,
+      "loss": 1.1072,
+      "step": 6400
+    },
+    {
+      "epoch": 1.588380716934487,
+      "grad_norm": 0.74609375,
+      "learning_rate": 6.961538461538462e-05,
+      "loss": 1.098,
+      "step": 6425
+    },
+    {
+      "epoch": 1.5945611866501854,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.94871794871795e-05,
+      "loss": 1.101,
+      "step": 6450
+    },
+    {
+      "epoch": 1.6007416563658838,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.935897435897436e-05,
+      "loss": 1.1008,
+      "step": 6475
+    },
+    {
+      "epoch": 1.6069221260815822,
+      "grad_norm": 0.796875,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 1.1,
+      "step": 6500
+    },
+    {
+      "epoch": 1.6131025957972804,
+      "grad_norm": 0.7265625,
+      "learning_rate": 6.91025641025641e-05,
+      "loss": 1.0997,
+      "step": 6525
+    },
+    {
+      "epoch": 1.6192830655129788,
+      "grad_norm": 0.7890625,
+      "learning_rate": 6.897435897435898e-05,
+      "loss": 1.1005,
+      "step": 6550
+    },
+    {
+      "epoch": 1.6254635352286773,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.884615384615385e-05,
+      "loss": 1.0953,
+      "step": 6575
+    },
+    {
+      "epoch": 1.6316440049443757,
+      "grad_norm": 0.71875,
+      "learning_rate": 6.871794871794872e-05,
+      "loss": 1.0999,
+      "step": 6600
+    },
+    {
+      "epoch": 1.637824474660074,
+      "grad_norm": 0.7265625,
+      "learning_rate": 6.858974358974359e-05,
+      "loss": 1.0965,
+      "step": 6625
+    },
+    {
+      "epoch": 1.6440049443757725,
+      "grad_norm": 0.73828125,
+      "learning_rate": 6.846153846153847e-05,
+      "loss": 1.098,
+      "step": 6650
+    },
+    {
+      "epoch": 1.650185414091471,
+      "grad_norm": 0.82421875,
+      "learning_rate": 6.833333333333333e-05,
+      "loss": 1.09,
+      "step": 6675
+    },
+    {
+      "epoch": 1.6563658838071693,
+      "grad_norm": 0.828125,
+      "learning_rate": 6.820512820512821e-05,
+      "loss": 1.0992,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6625463535228677,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.807692307692309e-05,
+      "loss": 1.0948,
+      "step": 6725
+    },
+    {
+      "epoch": 1.6687268232385661,
+      "grad_norm": 0.75,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 1.098,
+      "step": 6750
+    },
+    {
+      "epoch": 1.6749072929542645,
+      "grad_norm": 0.75,
+      "learning_rate": 6.782051282051283e-05,
+      "loss": 1.0915,
+      "step": 6775
+    },
+    {
+      "epoch": 1.681087762669963,
+      "grad_norm": 0.75,
+      "learning_rate": 6.76923076923077e-05,
+      "loss": 1.0943,
+      "step": 6800
+    },
+    {
+      "epoch": 1.6872682323856614,
+      "grad_norm": 0.6953125,
+      "learning_rate": 6.756410256410258e-05,
+      "loss": 1.0913,
+      "step": 6825
+    },
+    {
+      "epoch": 1.6934487021013598,
+      "grad_norm": 0.7265625,
+      "learning_rate": 6.743589743589744e-05,
+      "loss": 1.092,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6996291718170582,
+      "grad_norm": 0.8125,
+      "learning_rate": 6.730769230769232e-05,
+      "loss": 1.0904,
+      "step": 6875
+    },
+    {
+      "epoch": 1.7058096415327566,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.717948717948718e-05,
+      "loss": 1.091,
+      "step": 6900
+    },
+    {
+      "epoch": 1.711990111248455,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.705128205128205e-05,
+      "loss": 1.0901,
+      "step": 6925
+    },
+    {
+      "epoch": 1.7181705809641534,
+      "grad_norm": 0.765625,
+      "learning_rate": 6.692307692307693e-05,
+      "loss": 1.0885,
+      "step": 6950
+    },
+    {
+      "epoch": 1.7243510506798516,
+      "grad_norm": 0.74609375,
+      "learning_rate": 6.679487179487179e-05,
+      "loss": 1.0897,
+      "step": 6975
+    },
+    {
+      "epoch": 1.73053152039555,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.0888,
+      "step": 7000
+    },
+    {
+      "epoch": 1.7367119901112484,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.653846153846153e-05,
+      "loss": 1.0887,
+      "step": 7025
+    },
+    {
+      "epoch": 1.7428924598269468,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.641025641025641e-05,
+      "loss": 1.0857,
+      "step": 7050
+    },
+    {
+      "epoch": 1.7490729295426453,
+      "grad_norm": 0.8359375,
+      "learning_rate": 6.628205128205128e-05,
+      "loss": 1.089,
+      "step": 7075
+    },
+    {
+      "epoch": 1.7552533992583437,
+      "grad_norm": 0.75390625,
+      "learning_rate": 6.615384615384616e-05,
+      "loss": 1.0866,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7614338689740419,
+      "grad_norm": 0.73046875,
+      "learning_rate": 6.602564102564102e-05,
+      "loss": 1.0871,
+      "step": 7125
+    },
+    {
+      "epoch": 1.7676143386897403,
+      "grad_norm": 0.75,
+      "learning_rate": 6.58974358974359e-05,
+      "loss": 1.0779,
+      "step": 7150
+    },
+    {
+      "epoch": 1.7737948084054387,
+      "grad_norm": 0.89453125,
+      "learning_rate": 6.576923076923078e-05,
+      "loss": 1.0852,
+      "step": 7175
+    },
+    {
+      "epoch": 1.779975278121137,
+      "grad_norm": 0.7890625,
+      "learning_rate": 6.564102564102564e-05,
+      "loss": 1.0836,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7861557478368355,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.551282051282052e-05,
+      "loss": 1.0843,
+      "step": 7225
+    },
+    {
+      "epoch": 1.792336217552534,
+      "grad_norm": 0.7265625,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 1.0802,
+      "step": 7250
+    },
+    {
+      "epoch": 1.7985166872682323,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.525641025641026e-05,
+      "loss": 1.083,
+      "step": 7275
+    },
+    {
+      "epoch": 1.8046971569839307,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.512820512820513e-05,
+      "loss": 1.0798,
+      "step": 7300
+    },
+    {
+      "epoch": 1.8108776266996292,
+      "grad_norm": 0.72265625,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.0823,
+      "step": 7325
+    },
+    {
+      "epoch": 1.8170580964153276,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.487179487179487e-05,
+      "loss": 1.0814,
+      "step": 7350
+    },
+    {
+      "epoch": 1.823238566131026,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.474358974358975e-05,
+      "loss": 1.0827,
+      "step": 7375
+    },
+    {
+      "epoch": 1.8294190358467244,
+      "grad_norm": 0.71484375,
+      "learning_rate": 6.461538461538462e-05,
+      "loss": 1.0827,
+      "step": 7400
+    },
+    {
+      "epoch": 1.8355995055624228,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.44871794871795e-05,
+      "loss": 1.0799,
+      "step": 7425
+    },
+    {
+      "epoch": 1.8417799752781212,
+      "grad_norm": 0.71484375,
+      "learning_rate": 6.435897435897437e-05,
+      "loss": 1.0783,
+      "step": 7450
+    },
+    {
+      "epoch": 1.8479604449938196,
+      "grad_norm": 0.7890625,
+      "learning_rate": 6.423076923076924e-05,
+      "loss": 1.0758,
+      "step": 7475
+    },
+    {
+      "epoch": 1.854140914709518,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 1.078,
+      "step": 7500
+    },
+    {
+      "epoch": 1.8603213844252164,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.397435897435897e-05,
+      "loss": 1.0715,
+      "step": 7525
+    },
+    {
+      "epoch": 1.8665018541409149,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.384615384615385e-05,
+      "loss": 1.0761,
+      "step": 7550
+    },
+    {
+      "epoch": 1.8726823238566133,
+      "grad_norm": 0.7734375,
+      "learning_rate": 6.371794871794871e-05,
+      "loss": 1.076,
+      "step": 7575
+    },
+    {
+      "epoch": 1.8788627935723115,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.358974358974359e-05,
+      "loss": 1.0732,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8850432632880099,
+      "grad_norm": 0.76953125,
+      "learning_rate": 6.346153846153847e-05,
+      "loss": 1.073,
+      "step": 7625
+    },
+    {
+      "epoch": 1.8912237330037083,
+      "grad_norm": 0.75,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 1.07,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8974042027194067,
+      "grad_norm": 0.74609375,
+      "learning_rate": 6.320512820512821e-05,
+      "loss": 1.0739,
+      "step": 7675
+    },
+    {
+      "epoch": 1.903584672435105,
+      "grad_norm": 0.80078125,
+      "learning_rate": 6.307692307692308e-05,
+      "loss": 1.071,
+      "step": 7700
+    },
+    {
+      "epoch": 1.9097651421508035,
+      "grad_norm": 0.77734375,
+      "learning_rate": 6.294871794871795e-05,
+      "loss": 1.0706,
+      "step": 7725
+    },
+    {
+      "epoch": 1.9159456118665017,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 1.0725,
+      "step": 7750
+    },
+    {
+      "epoch": 1.9221260815822,
+      "grad_norm": 0.72265625,
+      "learning_rate": 6.26923076923077e-05,
+      "loss": 1.0666,
+      "step": 7775
+    },
+    {
+      "epoch": 1.9283065512978985,
+      "grad_norm": 0.7890625,
+      "learning_rate": 6.256410256410256e-05,
+      "loss": 1.0661,
+      "step": 7800
+    },
+    {
+      "epoch": 1.934487021013597,
+      "grad_norm": 0.71484375,
+      "learning_rate": 6.243589743589744e-05,
+      "loss": 1.0677,
+      "step": 7825
+    },
+    {
+      "epoch": 1.9406674907292953,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.23076923076923e-05,
+      "loss": 1.0683,
+      "step": 7850
+    },
+    {
+      "epoch": 1.9468479604449938,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.217948717948718e-05,
+      "loss": 1.0676,
+      "step": 7875
+    },
+    {
+      "epoch": 1.9530284301606922,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.205128205128206e-05,
+      "loss": 1.068,
+      "step": 7900
+    },
+    {
+      "epoch": 1.9592088998763906,
+      "grad_norm": 0.75390625,
+      "learning_rate": 6.192307692307693e-05,
+      "loss": 1.0662,
+      "step": 7925
+    },
+    {
+      "epoch": 1.965389369592089,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.17948717948718e-05,
+      "loss": 1.0653,
+      "step": 7950
+    },
+    {
+      "epoch": 1.9715698393077874,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.166666666666667e-05,
+      "loss": 1.0629,
+      "step": 7975
+    },
+    {
+      "epoch": 1.9777503090234858,
+      "grad_norm": 0.7265625,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 1.0669,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9839307787391842,
+      "grad_norm": 0.71875,
+      "learning_rate": 6.141025641025641e-05,
+      "loss": 1.0648,
+      "step": 8025
+    },
+    {
+      "epoch": 1.9901112484548826,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.128205128205129e-05,
+      "loss": 1.0646,
+      "step": 8050
+    },
+    {
+      "epoch": 1.996291718170581,
+      "grad_norm": 0.8125,
+      "learning_rate": 6.115384615384616e-05,
+      "loss": 1.0591,
+      "step": 8075
+    },
+    {
+      "epoch": 2.0024721878862795,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.1025641025641035e-05,
+      "loss": 1.0239,
+      "step": 8100
+    },
+    {
+      "epoch": 2.008652657601978,
+      "grad_norm": 0.73046875,
+      "learning_rate": 6.089743589743589e-05,
+      "loss": 0.9635,
+      "step": 8125
+    },
+    {
+      "epoch": 2.0148331273176763,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.0769230769230765e-05,
+      "loss": 0.9628,
+      "step": 8150
+    },
+    {
+      "epoch": 2.0210135970333747,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.0641025641025637e-05,
+      "loss": 0.9707,
+      "step": 8175
+    },
+    {
+      "epoch": 2.027194066749073,
+      "grad_norm": 0.78125,
+      "learning_rate": 6.0512820512820515e-05,
+      "loss": 0.9658,
+      "step": 8200
+    },
+    {
+      "epoch": 2.0333745364647715,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.038461538461539e-05,
+      "loss": 0.9695,
+      "step": 8225
+    },
+    {
+      "epoch": 2.03955500618047,
+      "grad_norm": 0.8046875,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 0.9714,
+      "step": 8250
+    },
+    {
+      "epoch": 2.0457354758961683,
+      "grad_norm": 0.78515625,
+      "learning_rate": 6.012820512820513e-05,
+      "loss": 0.973,
+      "step": 8275
+    },
+    {
+      "epoch": 2.0519159456118663,
+      "grad_norm": 0.80078125,
+      "learning_rate": 6e-05,
+      "loss": 0.9738,
+      "step": 8300
+    },
+    {
+      "epoch": 2.0580964153275647,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5.987179487179487e-05,
+      "loss": 0.9714,
+      "step": 8325
+    },
+    {
+      "epoch": 2.064276885043263,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.9743589743589745e-05,
+      "loss": 0.9718,
+      "step": 8350
+    },
+    {
+      "epoch": 2.0704573547589615,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.9615384615384616e-05,
+      "loss": 0.9768,
+      "step": 8375
+    },
+    {
+      "epoch": 2.07663782447466,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.948717948717949e-05,
+      "loss": 0.9756,
+      "step": 8400
+    },
+    {
+      "epoch": 2.0828182941903584,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.935897435897436e-05,
+      "loss": 0.9768,
+      "step": 8425
+    },
+    {
+      "epoch": 2.0889987639060568,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.923076923076923e-05,
+      "loss": 0.9755,
+      "step": 8450
+    },
+    {
+      "epoch": 2.095179233621755,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.910256410256411e-05,
+      "loss": 0.9758,
+      "step": 8475
+    },
+    {
+      "epoch": 2.1013597033374536,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.9762,
+      "step": 8500
+    },
+    {
+      "epoch": 2.107540173053152,
+      "grad_norm": 0.75,
+      "learning_rate": 5.884615384615385e-05,
+      "loss": 0.9774,
+      "step": 8525
+    },
+    {
+      "epoch": 2.1137206427688504,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.8717948717948725e-05,
+      "loss": 0.9795,
+      "step": 8550
+    },
+    {
+      "epoch": 2.119901112484549,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.8589743589743596e-05,
+      "loss": 0.9762,
+      "step": 8575
+    },
+    {
+      "epoch": 2.1260815822002472,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.846153846153847e-05,
+      "loss": 0.9808,
+      "step": 8600
+    },
+    {
+      "epoch": 2.1322620519159456,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.9781,
+      "step": 8625
+    },
+    {
+      "epoch": 2.138442521631644,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.820512820512821e-05,
+      "loss": 0.9772,
+      "step": 8650
+    },
+    {
+      "epoch": 2.1446229913473425,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.807692307692308e-05,
+      "loss": 0.9774,
+      "step": 8675
+    },
+    {
+      "epoch": 2.150803461063041,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.7948717948717954e-05,
+      "loss": 0.9764,
+      "step": 8700
+    },
+    {
+      "epoch": 2.1569839307787393,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.7820512820512826e-05,
+      "loss": 0.9781,
+      "step": 8725
+    },
+    {
+      "epoch": 2.1631644004944377,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.9748,
+      "step": 8750
+    },
+    {
+      "epoch": 2.169344870210136,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.756410256410256e-05,
+      "loss": 0.9762,
+      "step": 8775
+    },
+    {
+      "epoch": 2.1755253399258345,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.7435897435897434e-05,
+      "loss": 0.9789,
+      "step": 8800
+    },
+    {
+      "epoch": 2.181705809641533,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.7307692307692306e-05,
+      "loss": 0.9764,
+      "step": 8825
+    },
+    {
+      "epoch": 2.1878862793572313,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.717948717948718e-05,
+      "loss": 0.9767,
+      "step": 8850
+    },
+    {
+      "epoch": 2.1940667490729293,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.705128205128205e-05,
+      "loss": 0.9798,
+      "step": 8875
+    },
+    {
+      "epoch": 2.200247218788628,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.692307692307692e-05,
+      "loss": 0.9766,
+      "step": 8900
+    },
+    {
+      "epoch": 2.206427688504326,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5.679487179487179e-05,
+      "loss": 0.9814,
+      "step": 8925
+    },
+    {
+      "epoch": 2.2126081582200245,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.666666666666667e-05,
+      "loss": 0.9784,
+      "step": 8950
+    },
+    {
+      "epoch": 2.218788627935723,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5.653846153846154e-05,
+      "loss": 0.9749,
+      "step": 8975
+    },
+    {
+      "epoch": 2.2249690976514214,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.9814,
+      "step": 9000
+    },
+    {
+      "epoch": 2.23114956736712,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5.6282051282051286e-05,
+      "loss": 0.9805,
+      "step": 9025
+    },
+    {
+      "epoch": 2.237330037082818,
+      "grad_norm": 0.80859375,
+      "learning_rate": 5.615384615384616e-05,
+      "loss": 0.9749,
+      "step": 9050
+    },
+    {
+      "epoch": 2.2435105067985166,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.602564102564103e-05,
+      "loss": 0.9747,
+      "step": 9075
+    },
+    {
+      "epoch": 2.249690976514215,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.58974358974359e-05,
+      "loss": 0.9755,
+      "step": 9100
+    },
+    {
+      "epoch": 2.2558714462299134,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.576923076923077e-05,
+      "loss": 0.9767,
+      "step": 9125
+    },
+    {
+      "epoch": 2.262051915945612,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.5641025641025644e-05,
+      "loss": 0.9799,
+      "step": 9150
+    },
+    {
+      "epoch": 2.2682323856613102,
+      "grad_norm": 0.75390625,
+      "learning_rate": 5.5512820512820515e-05,
+      "loss": 0.9807,
+      "step": 9175
+    },
+    {
+      "epoch": 2.2744128553770087,
+      "grad_norm": 0.78125,
+      "learning_rate": 5.538461538461539e-05,
+      "loss": 0.9765,
+      "step": 9200
+    },
+    {
+      "epoch": 2.280593325092707,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.5256410256410265e-05,
+      "loss": 0.9767,
+      "step": 9225
+    },
+    {
+      "epoch": 2.2867737948084055,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 0.9751,
+      "step": 9250
+    },
+    {
+      "epoch": 2.292954264524104,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.9749,
+      "step": 9275
+    },
+    {
+      "epoch": 2.2991347342398023,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.487179487179488e-05,
+      "loss": 0.975,
+      "step": 9300
+    },
+    {
+      "epoch": 2.3053152039555007,
+      "grad_norm": 0.78125,
+      "learning_rate": 5.474358974358975e-05,
+      "loss": 0.9754,
+      "step": 9325
+    },
+    {
+      "epoch": 2.311495673671199,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.461538461538461e-05,
+      "loss": 0.9721,
+      "step": 9350
+    },
+    {
+      "epoch": 2.3176761433868975,
+      "grad_norm": 0.76953125,
+      "learning_rate": 5.448717948717948e-05,
+      "loss": 0.9739,
+      "step": 9375
+    },
+    {
+      "epoch": 2.323856613102596,
+      "grad_norm": 0.74609375,
+      "learning_rate": 5.435897435897436e-05,
+      "loss": 0.9759,
+      "step": 9400
+    },
+    {
+      "epoch": 2.3300370828182944,
+      "grad_norm": 0.75390625,
+      "learning_rate": 5.423076923076923e-05,
+      "loss": 0.9716,
+      "step": 9425
+    },
+    {
+      "epoch": 2.3362175525339928,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5.41025641025641e-05,
+      "loss": 0.9757,
+      "step": 9450
+    },
+    {
+      "epoch": 2.342398022249691,
+      "grad_norm": 0.73046875,
+      "learning_rate": 5.3974358974358975e-05,
+      "loss": 0.9723,
+      "step": 9475
+    },
+    {
+      "epoch": 2.348578491965389,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.973,
+      "step": 9500
+    },
+    {
+      "epoch": 2.354758961681088,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.371794871794872e-05,
+      "loss": 0.9721,
+      "step": 9525
+    },
+    {
+      "epoch": 2.360939431396786,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.358974358974359e-05,
+      "loss": 0.9762,
+      "step": 9550
+    },
+    {
+      "epoch": 2.3671199011124844,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.346153846153846e-05,
+      "loss": 0.9781,
+      "step": 9575
+    },
+    {
+      "epoch": 2.373300370828183,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.9767,
+      "step": 9600
+    },
+    {
+      "epoch": 2.379480840543881,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5.3205128205128205e-05,
+      "loss": 0.974,
+      "step": 9625
+    },
+    {
+      "epoch": 2.3856613102595796,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.3076923076923076e-05,
+      "loss": 0.9717,
+      "step": 9650
+    },
+    {
+      "epoch": 2.391841779975278,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.2948717948717955e-05,
+      "loss": 0.9717,
+      "step": 9675
+    },
+    {
+      "epoch": 2.3980222496909764,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.2820512820512826e-05,
+      "loss": 0.9726,
+      "step": 9700
+    },
+    {
+      "epoch": 2.404202719406675,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.26923076923077e-05,
+      "loss": 0.9746,
+      "step": 9725
+    },
+    {
+      "epoch": 2.4103831891223733,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 0.9739,
+      "step": 9750
+    },
+    {
+      "epoch": 2.4165636588380717,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.243589743589744e-05,
+      "loss": 0.9702,
+      "step": 9775
+    },
+    {
+      "epoch": 2.42274412855377,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.230769230769231e-05,
+      "loss": 0.97,
+      "step": 9800
+    },
+    {
+      "epoch": 2.4289245982694685,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.2179487179487185e-05,
+      "loss": 0.9698,
+      "step": 9825
+    },
+    {
+      "epoch": 2.435105067985167,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.2051282051282056e-05,
+      "loss": 0.9687,
+      "step": 9850
+    },
+    {
+      "epoch": 2.4412855377008653,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.192307692307693e-05,
+      "loss": 0.9673,
+      "step": 9875
+    },
+    {
+      "epoch": 2.4474660074165637,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.17948717948718e-05,
+      "loss": 0.9671,
+      "step": 9900
+    },
+    {
+      "epoch": 2.453646477132262,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.166666666666667e-05,
+      "loss": 0.9685,
+      "step": 9925
+    },
+    {
+      "epoch": 2.4598269468479605,
+      "grad_norm": 0.83203125,
+      "learning_rate": 5.1538461538461536e-05,
+      "loss": 0.9722,
+      "step": 9950
+    },
+    {
+      "epoch": 2.466007416563659,
+      "grad_norm": 0.875,
+      "learning_rate": 5.141025641025641e-05,
+      "loss": 0.969,
+      "step": 9975
+    },
+    {
+      "epoch": 2.4721878862793574,
+      "grad_norm": 0.80078125,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.9717,
+      "step": 10000
+    },
+    {
+      "epoch": 2.4721878862793574,
+      "eval_loss": 1.0057789087295532,
+      "eval_runtime": 1.5251,
+      "eval_samples_per_second": 418.982,
+      "eval_steps_per_second": 1.967,
+      "step": 10000
+    },
+    {
+      "epoch": 2.478368355995056,
+      "grad_norm": 0.765625,
+      "learning_rate": 5.115384615384615e-05,
+      "loss": 0.9694,
+      "step": 10025
+    },
+    {
+      "epoch": 2.484548825710754,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.102564102564102e-05,
+      "loss": 0.9696,
+      "step": 10050
+    },
+    {
+      "epoch": 2.490729295426452,
+      "grad_norm": 0.734375,
+      "learning_rate": 5.0897435897435894e-05,
+      "loss": 0.9687,
+      "step": 10075
+    },
+    {
+      "epoch": 2.496909765142151,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5.0769230769230766e-05,
+      "loss": 0.964,
+      "step": 10100
+    },
+    {
+      "epoch": 2.503090234857849,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.0641025641025644e-05,
+      "loss": 0.9691,
+      "step": 10125
+    },
+    {
+      "epoch": 2.509270704573548,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5.0512820512820516e-05,
+      "loss": 0.9649,
+      "step": 10150
+    },
+    {
+      "epoch": 2.515451174289246,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.038461538461539e-05,
+      "loss": 0.9673,
+      "step": 10175
+    },
+    {
+      "epoch": 2.521631644004944,
+      "grad_norm": 0.75390625,
+      "learning_rate": 5.025641025641026e-05,
+      "loss": 0.9669,
+      "step": 10200
+    },
+    {
+      "epoch": 2.5278121137206426,
+      "grad_norm": 0.765625,
+      "learning_rate": 5.012820512820513e-05,
+      "loss": 0.9653,
+      "step": 10225
+    },
+    {
+      "epoch": 2.533992583436341,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5e-05,
+      "loss": 0.9682,
+      "step": 10250
+    },
+    {
+      "epoch": 2.5401730531520395,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.9871794871794874e-05,
+      "loss": 0.9615,
+      "step": 10275
+    },
+    {
+      "epoch": 2.546353522867738,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9743589743589746e-05,
+      "loss": 0.9618,
+      "step": 10300
+    },
+    {
+      "epoch": 2.5525339925834363,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.961538461538462e-05,
+      "loss": 0.9633,
+      "step": 10325
+    },
+    {
+      "epoch": 2.5587144622991347,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.948717948717949e-05,
+      "loss": 0.9624,
+      "step": 10350
+    },
+    {
+      "epoch": 2.564894932014833,
+      "grad_norm": 0.75,
+      "learning_rate": 4.935897435897436e-05,
+      "loss": 0.9645,
+      "step": 10375
+    },
+    {
+      "epoch": 2.5710754017305315,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.923076923076924e-05,
+      "loss": 0.9594,
+      "step": 10400
+    },
+    {
+      "epoch": 2.57725587144623,
+      "grad_norm": 0.70703125,
+      "learning_rate": 4.9102564102564104e-05,
+      "loss": 0.9652,
+      "step": 10425
+    },
+    {
+      "epoch": 2.5834363411619283,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.8974358974358975e-05,
+      "loss": 0.9606,
+      "step": 10450
+    },
+    {
+      "epoch": 2.5896168108776267,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.884615384615385e-05,
+      "loss": 0.9639,
+      "step": 10475
+    },
+    {
+      "epoch": 2.595797280593325,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.9615,
+      "step": 10500
+    },
+    {
+      "epoch": 2.6019777503090236,
+      "grad_norm": 0.80078125,
+      "learning_rate": 4.858974358974359e-05,
+      "loss": 0.961,
+      "step": 10525
+    },
+    {
+      "epoch": 2.608158220024722,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.846153846153846e-05,
+      "loss": 0.9608,
+      "step": 10550
+    },
+    {
+      "epoch": 2.6143386897404204,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.9605,
+      "step": 10575
+    },
+    {
+      "epoch": 2.620519159456119,
+      "grad_norm": 0.7421875,
+      "learning_rate": 4.8205128205128205e-05,
+      "loss": 0.9572,
+      "step": 10600
+    },
+    {
+      "epoch": 2.626699629171817,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 0.961,
+      "step": 10625
+    },
+    {
+      "epoch": 2.632880098887515,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.7948717948717955e-05,
+      "loss": 0.9538,
+      "step": 10650
+    },
+    {
+      "epoch": 2.639060568603214,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.782051282051283e-05,
+      "loss": 0.9529,
+      "step": 10675
+    },
+    {
+      "epoch": 2.645241038318912,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.76923076923077e-05,
+      "loss": 0.9579,
+      "step": 10700
+    },
+    {
+      "epoch": 2.651421508034611,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.7564102564102563e-05,
+      "loss": 0.956,
+      "step": 10725
+    },
+    {
+      "epoch": 2.657601977750309,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 0.9592,
+      "step": 10750
+    },
+    {
+      "epoch": 2.6637824474660077,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.730769230769231e-05,
+      "loss": 0.9578,
+      "step": 10775
+    },
+    {
+      "epoch": 2.6699629171817056,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.717948717948718e-05,
+      "loss": 0.9579,
+      "step": 10800
+    },
+    {
+      "epoch": 2.676143386897404,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.705128205128205e-05,
+      "loss": 0.9561,
+      "step": 10825
+    },
+    {
+      "epoch": 2.6823238566131025,
+      "grad_norm": 0.70703125,
+      "learning_rate": 4.692307692307693e-05,
+      "loss": 0.9567,
+      "step": 10850
+    },
+    {
+      "epoch": 2.688504326328801,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.67948717948718e-05,
+      "loss": 0.956,
+      "step": 10875
+    },
+    {
+      "epoch": 2.6946847960444993,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.9567,
+      "step": 10900
+    },
+    {
+      "epoch": 2.7008652657601977,
+      "grad_norm": 0.80078125,
+      "learning_rate": 4.653846153846154e-05,
+      "loss": 0.9514,
+      "step": 10925
+    },
+    {
+      "epoch": 2.707045735475896,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.6410256410256415e-05,
+      "loss": 0.9535,
+      "step": 10950
+    },
+    {
+      "epoch": 2.7132262051915945,
+      "grad_norm": 0.87890625,
+      "learning_rate": 4.6282051282051287e-05,
+      "loss": 0.9571,
+      "step": 10975
+    },
+    {
+      "epoch": 2.719406674907293,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.9511,
+      "step": 11000
+    },
+    {
+      "epoch": 2.7255871446229913,
+      "grad_norm": 0.8046875,
+      "learning_rate": 4.602564102564102e-05,
+      "loss": 0.9511,
+      "step": 11025
+    },
+    {
+      "epoch": 2.7317676143386898,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.5897435897435895e-05,
+      "loss": 0.9554,
+      "step": 11050
+    },
+    {
+      "epoch": 2.737948084054388,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.576923076923077e-05,
+      "loss": 0.95,
+      "step": 11075
+    },
+    {
+      "epoch": 2.7441285537700866,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.5641025641025645e-05,
+      "loss": 0.951,
+      "step": 11100
+    },
+    {
+      "epoch": 2.750309023485785,
+      "grad_norm": 0.71875,
+      "learning_rate": 4.5512820512820516e-05,
+      "loss": 0.9475,
+      "step": 11125
+    },
+    {
+      "epoch": 2.7564894932014834,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.538461538461539e-05,
+      "loss": 0.9511,
+      "step": 11150
+    },
+    {
+      "epoch": 2.762669962917182,
+      "grad_norm": 0.75,
+      "learning_rate": 4.525641025641026e-05,
+      "loss": 0.9489,
+      "step": 11175
+    },
+    {
+      "epoch": 2.76885043263288,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.512820512820513e-05,
+      "loss": 0.9511,
+      "step": 11200
+    },
+    {
+      "epoch": 2.7750309023485786,
+      "grad_norm": 0.734375,
+      "learning_rate": 4.5e-05,
+      "loss": 0.951,
+      "step": 11225
+    },
+    {
+      "epoch": 2.781211372064277,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 0.9532,
+      "step": 11250
+    },
+    {
+      "epoch": 2.787391841779975,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.4743589743589746e-05,
+      "loss": 0.9517,
+      "step": 11275
+    },
+    {
+      "epoch": 2.793572311495674,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.461538461538462e-05,
+      "loss": 0.9479,
+      "step": 11300
+    },
+    {
+      "epoch": 2.799752781211372,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.448717948717949e-05,
+      "loss": 0.9493,
+      "step": 11325
+    },
+    {
+      "epoch": 2.8059332509270707,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.435897435897436e-05,
+      "loss": 0.9489,
+      "step": 11350
+    },
+    {
+      "epoch": 2.8121137206427687,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.423076923076923e-05,
+      "loss": 0.945,
+      "step": 11375
+    },
+    {
+      "epoch": 2.8182941903584675,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.4102564102564104e-05,
+      "loss": 0.9511,
+      "step": 11400
+    },
+    {
+      "epoch": 2.8244746600741655,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.3974358974358976e-05,
+      "loss": 0.9455,
+      "step": 11425
+    },
+    {
+      "epoch": 2.830655129789864,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.384615384615385e-05,
+      "loss": 0.9468,
+      "step": 11450
+    },
+    {
+      "epoch": 2.8368355995055623,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.371794871794872e-05,
+      "loss": 0.9478,
+      "step": 11475
+    },
+    {
+      "epoch": 2.8430160692212607,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.9448,
+      "step": 11500
+    },
+    {
+      "epoch": 2.849196538936959,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.346153846153846e-05,
+      "loss": 0.9439,
+      "step": 11525
+    },
+    {
+      "epoch": 2.8553770086526575,
+      "grad_norm": 0.703125,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.948,
+      "step": 11550
+    },
+    {
+      "epoch": 2.861557478368356,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.320512820512821e-05,
+      "loss": 0.9465,
+      "step": 11575
+    },
+    {
+      "epoch": 2.8677379480840544,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.3076923076923084e-05,
+      "loss": 0.9469,
+      "step": 11600
+    },
+    {
+      "epoch": 2.8739184177997528,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.294871794871795e-05,
+      "loss": 0.9417,
+      "step": 11625
+    },
+    {
+      "epoch": 2.880098887515451,
+      "grad_norm": 0.6953125,
+      "learning_rate": 4.282051282051282e-05,
+      "loss": 0.9431,
+      "step": 11650
+    },
+    {
+      "epoch": 2.8862793572311496,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.269230769230769e-05,
+      "loss": 0.9446,
+      "step": 11675
+    },
+    {
+      "epoch": 2.892459826946848,
+      "grad_norm": 0.734375,
+      "learning_rate": 4.2564102564102564e-05,
+      "loss": 0.943,
+      "step": 11700
+    },
+    {
+      "epoch": 2.8986402966625464,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.2435897435897435e-05,
+      "loss": 0.9467,
+      "step": 11725
+    },
+    {
+      "epoch": 2.904820766378245,
+      "grad_norm": 0.7265625,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.9489,
+      "step": 11750
+    },
+    {
+      "epoch": 2.9110012360939432,
+      "grad_norm": 0.72265625,
+      "learning_rate": 4.217948717948718e-05,
+      "loss": 0.9435,
+      "step": 11775
+    },
+    {
+      "epoch": 2.9171817058096416,
+      "grad_norm": 0.734375,
+      "learning_rate": 4.205128205128206e-05,
+      "loss": 0.9405,
+      "step": 11800
+    },
+    {
+      "epoch": 2.92336217552534,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.192307692307693e-05,
+      "loss": 0.9422,
+      "step": 11825
+    },
+    {
+      "epoch": 2.9295426452410385,
+      "grad_norm": 0.73828125,
+      "learning_rate": 4.17948717948718e-05,
+      "loss": 0.9421,
+      "step": 11850
+    },
+    {
+      "epoch": 2.935723114956737,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.9394,
+      "step": 11875
+    },
+    {
+      "epoch": 2.941903584672435,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.1538461538461544e-05,
+      "loss": 0.9416,
+      "step": 11900
+    },
+    {
+      "epoch": 2.9480840543881337,
+      "grad_norm": 0.71484375,
+      "learning_rate": 4.1410256410256415e-05,
+      "loss": 0.9412,
+      "step": 11925
+    },
+    {
+      "epoch": 2.9542645241038317,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.128205128205128e-05,
+      "loss": 0.9399,
+      "step": 11950
+    },
+    {
+      "epoch": 2.9604449938195305,
+      "grad_norm": 0.70703125,
+      "learning_rate": 4.115384615384615e-05,
+      "loss": 0.9396,
+      "step": 11975
+    },
+    {
+      "epoch": 2.9666254635352285,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.9366,
+      "step": 12000
+    },
+    {
+      "epoch": 2.9728059332509273,
+      "grad_norm": 0.73828125,
+      "learning_rate": 4.0897435897435895e-05,
+      "loss": 0.9368,
+      "step": 12025
+    },
+    {
+      "epoch": 2.9789864029666253,
+      "grad_norm": 0.72265625,
+      "learning_rate": 4.0769230769230773e-05,
+      "loss": 0.9392,
+      "step": 12050
+    },
+    {
+      "epoch": 2.9851668726823237,
+      "grad_norm": 0.77734375,
+      "learning_rate": 4.0641025641025645e-05,
+      "loss": 0.9397,
+      "step": 12075
+    },
+    {
+      "epoch": 2.991347342398022,
+      "grad_norm": 0.7421875,
+      "learning_rate": 4.051282051282052e-05,
+      "loss": 0.9385,
+      "step": 12100
+    },
+    {
+      "epoch": 2.9975278121137205,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.038461538461539e-05,
+      "loss": 0.9397,
+      "step": 12125
+    },
+    {
+      "epoch": 3.003708281829419,
+      "grad_norm": 0.84375,
+      "learning_rate": 4.025641025641026e-05,
+      "loss": 0.8884,
+      "step": 12150
+    },
+    {
+      "epoch": 3.0098887515451174,
+      "grad_norm": 0.75,
+      "learning_rate": 4.012820512820513e-05,
+      "loss": 0.8596,
+      "step": 12175
+    },
+    {
+      "epoch": 3.016069221260816,
+      "grad_norm": 0.765625,
+      "learning_rate": 4e-05,
+      "loss": 0.8606,
+      "step": 12200
+    },
+    {
+      "epoch": 3.022249690976514,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.9871794871794875e-05,
+      "loss": 0.8623,
+      "step": 12225
+    },
+    {
+      "epoch": 3.0284301606922126,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 0.8606,
+      "step": 12250
+    },
+    {
+      "epoch": 3.034610630407911,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.961538461538462e-05,
+      "loss": 0.8626,
+      "step": 12275
+    },
+    {
+      "epoch": 3.0407911001236094,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.948717948717949e-05,
+      "loss": 0.8634,
+      "step": 12300
+    },
+    {
+      "epoch": 3.046971569839308,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.935897435897436e-05,
+      "loss": 0.8615,
+      "step": 12325
+    },
+    {
+      "epoch": 3.0531520395550062,
+      "grad_norm": 0.75,
+      "learning_rate": 3.923076923076923e-05,
+      "loss": 0.8622,
+      "step": 12350
+    },
+    {
+      "epoch": 3.0593325092707047,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.9102564102564105e-05,
+      "loss": 0.8627,
+      "step": 12375
+    },
+    {
+      "epoch": 3.065512978986403,
+      "grad_norm": 0.7578125,
+      "learning_rate": 3.8974358974358976e-05,
+      "loss": 0.8603,
+      "step": 12400
+    },
+    {
+      "epoch": 3.0716934487021015,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.884615384615385e-05,
+      "loss": 0.8676,
+      "step": 12425
+    },
+    {
+      "epoch": 3.0778739184178,
+      "grad_norm": 0.7421875,
+      "learning_rate": 3.871794871794872e-05,
+      "loss": 0.8622,
+      "step": 12450
+    },
+    {
+      "epoch": 3.0840543881334983,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.858974358974359e-05,
+      "loss": 0.8625,
+      "step": 12475
+    },
+    {
+      "epoch": 3.0902348578491967,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.8649,
+      "step": 12500
+    },
+    {
+      "epoch": 3.096415327564895,
+      "grad_norm": 0.76953125,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.8663,
+      "step": 12525
+    },
+    {
+      "epoch": 3.1025957972805935,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.8205128205128206e-05,
+      "loss": 0.8633,
+      "step": 12550
+    },
+    {
+      "epoch": 3.1087762669962915,
+      "grad_norm": 0.7578125,
+      "learning_rate": 3.807692307692308e-05,
+      "loss": 0.8652,
+      "step": 12575
+    },
+    {
+      "epoch": 3.11495673671199,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.794871794871795e-05,
+      "loss": 0.8671,
+      "step": 12600
+    },
+    {
+      "epoch": 3.1211372064276883,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.782051282051282e-05,
+      "loss": 0.8633,
+      "step": 12625
+    },
+    {
+      "epoch": 3.1273176761433867,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.769230769230769e-05,
+      "loss": 0.8672,
+      "step": 12650
+    },
+    {
+      "epoch": 3.133498145859085,
+      "grad_norm": 0.7421875,
+      "learning_rate": 3.7564102564102564e-05,
+      "loss": 0.868,
+      "step": 12675
+    },
+    {
+      "epoch": 3.1396786155747836,
+      "grad_norm": 0.734375,
+      "learning_rate": 3.7435897435897436e-05,
+      "loss": 0.8656,
+      "step": 12700
+    },
+    {
+      "epoch": 3.145859085290482,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.730769230769231e-05,
+      "loss": 0.8663,
+      "step": 12725
+    },
+    {
+      "epoch": 3.1520395550061804,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 0.8679,
+      "step": 12750
+    },
+    {
+      "epoch": 3.158220024721879,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.705128205128206e-05,
+      "loss": 0.867,
+      "step": 12775
+    },
+    {
+      "epoch": 3.164400494437577,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 0.8685,
+      "step": 12800
+    },
+    {
+      "epoch": 3.1705809641532756,
+      "grad_norm": 0.76953125,
+      "learning_rate": 3.67948717948718e-05,
+      "loss": 0.8673,
+      "step": 12825
+    },
+    {
+      "epoch": 3.176761433868974,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.8688,
+      "step": 12850
+    },
+    {
+      "epoch": 3.1829419035846724,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.653846153846154e-05,
+      "loss": 0.8638,
+      "step": 12875
+    },
+    {
+      "epoch": 3.189122373300371,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.641025641025641e-05,
+      "loss": 0.87,
+      "step": 12900
+    },
+    {
+      "epoch": 3.1953028430160693,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.628205128205128e-05,
+      "loss": 0.8717,
+      "step": 12925
+    },
+    {
+      "epoch": 3.2014833127317677,
+      "grad_norm": 0.734375,
+      "learning_rate": 3.615384615384615e-05,
+      "loss": 0.863,
+      "step": 12950
+    },
+    {
+      "epoch": 3.207663782447466,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.6025641025641024e-05,
+      "loss": 0.8704,
+      "step": 12975
+    },
+    {
+      "epoch": 3.2138442521631645,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 0.8695,
+      "step": 13000
+    },
+    {
+      "epoch": 3.220024721878863,
+      "grad_norm": 0.73046875,
+      "learning_rate": 3.5769230769230774e-05,
+      "loss": 0.8695,
+      "step": 13025
+    },
+    {
+      "epoch": 3.2262051915945613,
+      "grad_norm": 0.734375,
+      "learning_rate": 3.5641025641025646e-05,
+      "loss": 0.8668,
+      "step": 13050
+    },
+    {
+      "epoch": 3.2323856613102597,
+      "grad_norm": 0.77734375,
+      "learning_rate": 3.551282051282052e-05,
+      "loss": 0.8703,
+      "step": 13075
+    },
+    {
+      "epoch": 3.238566131025958,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.538461538461539e-05,
+      "loss": 0.8671,
+      "step": 13100
+    },
+    {
+      "epoch": 3.2447466007416566,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.525641025641026e-05,
+      "loss": 0.8677,
+      "step": 13125
+    },
+    {
+      "epoch": 3.2509270704573545,
+      "grad_norm": 0.7109375,
+      "learning_rate": 3.5128205128205125e-05,
+      "loss": 0.8671,
+      "step": 13150
+    },
+    {
+      "epoch": 3.2571075401730534,
+      "grad_norm": 0.71875,
+      "learning_rate": 3.5e-05,
+      "loss": 0.8682,
+      "step": 13175
+    },
+    {
+      "epoch": 3.2632880098887513,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.487179487179487e-05,
+      "loss": 0.8677,
+      "step": 13200
+    },
+    {
+      "epoch": 3.2694684796044498,
+      "grad_norm": 0.703125,
+      "learning_rate": 3.474358974358975e-05,
+      "loss": 0.871,
+      "step": 13225
+    },
+    {
+      "epoch": 3.275648949320148,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 0.8682,
+      "step": 13250
+    },
+    {
+      "epoch": 3.2818294190358466,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.448717948717949e-05,
+      "loss": 0.8653,
+      "step": 13275
+    },
+    {
+      "epoch": 3.288009888751545,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.435897435897436e-05,
+      "loss": 0.8697,
+      "step": 13300
+    },
+    {
+      "epoch": 3.2941903584672434,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.4230769230769234e-05,
+      "loss": 0.8697,
+      "step": 13325
+    },
+    {
+      "epoch": 3.300370828182942,
+      "grad_norm": 0.7578125,
+      "learning_rate": 3.4102564102564105e-05,
+      "loss": 0.869,
+      "step": 13350
+    },
+    {
+      "epoch": 3.30655129789864,
+      "grad_norm": 0.73046875,
+      "learning_rate": 3.397435897435898e-05,
+      "loss": 0.8687,
+      "step": 13375
+    },
+    {
+      "epoch": 3.3127317676143386,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 0.8692,
+      "step": 13400
+    },
+    {
+      "epoch": 3.318912237330037,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.371794871794872e-05,
+      "loss": 0.8687,
+      "step": 13425
+    },
+    {
+      "epoch": 3.3250927070457355,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.358974358974359e-05,
+      "loss": 0.8708,
+      "step": 13450
+    },
+    {
+      "epoch": 3.331273176761434,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.346153846153846e-05,
+      "loss": 0.8659,
+      "step": 13475
+    },
+    {
+      "epoch": 3.3374536464771323,
+      "grad_norm": 0.7578125,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.8662,
+      "step": 13500
+    },
+    {
+      "epoch": 3.3436341161928307,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.3205128205128207e-05,
+      "loss": 0.8692,
+      "step": 13525
+    },
+    {
+      "epoch": 3.349814585908529,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.307692307692308e-05,
+      "loss": 0.8677,
+      "step": 13550
+    },
+    {
+      "epoch": 3.3559950556242275,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.294871794871795e-05,
+      "loss": 0.8672,
+      "step": 13575
+    },
+    {
+      "epoch": 3.362175525339926,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.282051282051282e-05,
+      "loss": 0.8703,
+      "step": 13600
+    },
+    {
+      "epoch": 3.3683559950556243,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.269230769230769e-05,
+      "loss": 0.8658,
+      "step": 13625
+    },
+    {
+      "epoch": 3.3745364647713227,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.2564102564102565e-05,
+      "loss": 0.8679,
+      "step": 13650
+    },
+    {
+      "epoch": 3.380716934487021,
+      "grad_norm": 0.76953125,
+      "learning_rate": 3.2435897435897436e-05,
+      "loss": 0.8654,
+      "step": 13675
+    },
+    {
+      "epoch": 3.3868974042027196,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 0.8662,
+      "step": 13700
+    },
+    {
+      "epoch": 3.393077873918418,
+      "grad_norm": 0.73046875,
+      "learning_rate": 3.2179487179487186e-05,
+      "loss": 0.8654,
+      "step": 13725
+    },
+    {
+      "epoch": 3.3992583436341164,
+      "grad_norm": 0.734375,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 0.8696,
+      "step": 13750
+    },
+    {
+      "epoch": 3.4054388133498144,
+      "grad_norm": 0.7421875,
+      "learning_rate": 3.192307692307692e-05,
+      "loss": 0.8648,
+      "step": 13775
+    },
+    {
+      "epoch": 3.411619283065513,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.1794871794871795e-05,
+      "loss": 0.8677,
+      "step": 13800
+    },
+    {
+      "epoch": 3.417799752781211,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.8665,
+      "step": 13825
+    },
+    {
+      "epoch": 3.4239802224969096,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.153846153846154e-05,
+      "loss": 0.8662,
+      "step": 13850
+    },
+    {
+      "epoch": 3.430160692212608,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.141025641025641e-05,
+      "loss": 0.8658,
+      "step": 13875
+    },
+    {
+      "epoch": 3.4363411619283064,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.128205128205128e-05,
+      "loss": 0.8652,
+      "step": 13900
+    },
+    {
+      "epoch": 3.442521631644005,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.115384615384615e-05,
+      "loss": 0.8667,
+      "step": 13925
+    },
+    {
+      "epoch": 3.4487021013597032,
+      "grad_norm": 0.73046875,
+      "learning_rate": 3.102564102564103e-05,
+      "loss": 0.8654,
+      "step": 13950
+    },
+    {
+      "epoch": 3.4548825710754016,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.08974358974359e-05,
+      "loss": 0.8666,
+      "step": 13975
+    },
+    {
+      "epoch": 3.4610630407911,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.8712,
+      "step": 14000
+    },
+    {
+      "epoch": 3.4672435105067985,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.0641025641025646e-05,
+      "loss": 0.867,
+      "step": 14025
+    },
+    {
+      "epoch": 3.473423980222497,
+      "grad_norm": 0.72265625,
+      "learning_rate": 3.0512820512820518e-05,
+      "loss": 0.8635,
+      "step": 14050
+    },
+    {
+      "epoch": 3.4796044499381953,
+      "grad_norm": 0.765625,
+      "learning_rate": 3.0384615384615382e-05,
+      "loss": 0.8686,
+      "step": 14075
+    },
+    {
+      "epoch": 3.4857849196538937,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.0256410256410257e-05,
+      "loss": 0.8654,
+      "step": 14100
+    },
+    {
+      "epoch": 3.491965389369592,
+      "grad_norm": 0.71875,
+      "learning_rate": 3.012820512820513e-05,
+      "loss": 0.869,
+      "step": 14125
+    },
+    {
+      "epoch": 3.4981458590852905,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3e-05,
+      "loss": 0.862,
+      "step": 14150
+    },
+    {
+      "epoch": 3.504326328800989,
+      "grad_norm": 0.76171875,
+      "learning_rate": 2.9871794871794872e-05,
+      "loss": 0.866,
+      "step": 14175
+    },
+    {
+      "epoch": 3.5105067985166873,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.9743589743589744e-05,
+      "loss": 0.8701,
+      "step": 14200
+    },
+    {
+      "epoch": 3.5166872682323858,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.9615384615384616e-05,
+      "loss": 0.8679,
+      "step": 14225
+    },
+    {
+      "epoch": 3.522867737948084,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 0.8639,
+      "step": 14250
+    },
+    {
+      "epoch": 3.5290482076637826,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.9358974358974362e-05,
+      "loss": 0.8674,
+      "step": 14275
+    },
+    {
+      "epoch": 3.535228677379481,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.9230769230769234e-05,
+      "loss": 0.8639,
+      "step": 14300
+    },
+    {
+      "epoch": 3.5414091470951794,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.9102564102564106e-05,
+      "loss": 0.8652,
+      "step": 14325
+    },
+    {
+      "epoch": 3.5475896168108774,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.8974358974358977e-05,
+      "loss": 0.8687,
+      "step": 14350
+    },
+    {
+      "epoch": 3.553770086526576,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 0.8664,
+      "step": 14375
+    },
+    {
+      "epoch": 3.559950556242274,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.8717948717948717e-05,
+      "loss": 0.8671,
+      "step": 14400
+    },
+    {
+      "epoch": 3.566131025957973,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.858974358974359e-05,
+      "loss": 0.8658,
+      "step": 14425
+    },
+    {
+      "epoch": 3.572311495673671,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.846153846153846e-05,
+      "loss": 0.8598,
+      "step": 14450
+    },
+    {
+      "epoch": 3.57849196538937,
+      "grad_norm": 0.74609375,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.8643,
+      "step": 14475
+    },
+    {
+      "epoch": 3.584672435105068,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 0.8641,
+      "step": 14500
+    },
+    {
+      "epoch": 3.5908529048207662,
+      "grad_norm": 0.75,
+      "learning_rate": 2.807692307692308e-05,
+      "loss": 0.8665,
+      "step": 14525
+    },
+    {
+      "epoch": 3.5970333745364647,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.794871794871795e-05,
+      "loss": 0.8649,
+      "step": 14550
+    },
+    {
+      "epoch": 3.603213844252163,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.7820512820512822e-05,
+      "loss": 0.8671,
+      "step": 14575
+    },
+    {
+      "epoch": 3.6093943139678615,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 0.8656,
+      "step": 14600
+    },
+    {
+      "epoch": 3.61557478368356,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.756410256410257e-05,
+      "loss": 0.8642,
+      "step": 14625
+    },
+    {
+      "epoch": 3.6217552533992583,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.743589743589744e-05,
+      "loss": 0.8628,
+      "step": 14650
+    },
+    {
+      "epoch": 3.6279357231149567,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.7307692307692305e-05,
+      "loss": 0.8648,
+      "step": 14675
+    },
+    {
+      "epoch": 3.634116192830655,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.717948717948718e-05,
+      "loss": 0.8674,
+      "step": 14700
+    },
+    {
+      "epoch": 3.6402966625463535,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.705128205128205e-05,
+      "loss": 0.8664,
+      "step": 14725
+    },
+    {
+      "epoch": 3.646477132262052,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.8643,
+      "step": 14750
+    },
+    {
+      "epoch": 3.6526576019777504,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.6794871794871795e-05,
+      "loss": 0.8679,
+      "step": 14775
+    },
+    {
+      "epoch": 3.6588380716934488,
+      "grad_norm": 0.76171875,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.8669,
+      "step": 14800
+    },
+    {
+      "epoch": 3.665018541409147,
+      "grad_norm": 0.74609375,
+      "learning_rate": 2.6538461538461538e-05,
+      "loss": 0.8619,
+      "step": 14825
+    },
+    {
+      "epoch": 3.6711990111248456,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.6410256410256413e-05,
+      "loss": 0.8685,
+      "step": 14850
+    },
+    {
+      "epoch": 3.677379480840544,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.6282051282051285e-05,
+      "loss": 0.8666,
+      "step": 14875
+    },
+    {
+      "epoch": 3.6835599505562424,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.6153846153846157e-05,
+      "loss": 0.8645,
+      "step": 14900
+    },
+    {
+      "epoch": 3.689740420271941,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.6025641025641028e-05,
+      "loss": 0.8657,
+      "step": 14925
+    },
+    {
+      "epoch": 3.6959208899876392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 2.58974358974359e-05,
+      "loss": 0.8611,
+      "step": 14950
+    },
+    {
+      "epoch": 3.702101359703337,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.5769230769230768e-05,
+      "loss": 0.8628,
+      "step": 14975
+    },
+    {
+      "epoch": 3.708281829419036,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 0.8643,
+      "step": 15000
+    },
+    {
+      "epoch": 3.708281829419036,
+      "eval_loss": 0.9966387748718262,
+      "eval_runtime": 1.5383,
+      "eval_samples_per_second": 415.389,
+      "eval_steps_per_second": 1.95,
+      "step": 15000
+    },
+    {
+      "epoch": 3.714462299134734,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.551282051282051e-05,
+      "loss": 0.864,
+      "step": 15025
+    },
+    {
+      "epoch": 3.720642768850433,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.5384615384615383e-05,
+      "loss": 0.8643,
+      "step": 15050
+    },
+    {
+      "epoch": 3.726823238566131,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.5256410256410258e-05,
+      "loss": 0.8614,
+      "step": 15075
+    },
+    {
+      "epoch": 3.7330037082818293,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.512820512820513e-05,
+      "loss": 0.8694,
+      "step": 15100
+    },
+    {
+      "epoch": 3.7391841779975277,
+      "grad_norm": 0.75,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8666,
+      "step": 15125
+    },
+    {
+      "epoch": 3.745364647713226,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.4871794871794873e-05,
+      "loss": 0.8616,
+      "step": 15150
+    },
+    {
+      "epoch": 3.7515451174289245,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.4743589743589744e-05,
+      "loss": 0.8646,
+      "step": 15175
+    },
+    {
+      "epoch": 3.757725587144623,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 0.8623,
+      "step": 15200
+    },
+    {
+      "epoch": 3.7639060568603213,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.4487179487179488e-05,
+      "loss": 0.8645,
+      "step": 15225
+    },
+    {
+      "epoch": 3.7700865265760197,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 0.8612,
+      "step": 15250
+    },
+    {
+      "epoch": 3.776266996291718,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.423076923076923e-05,
+      "loss": 0.8642,
+      "step": 15275
+    },
+    {
+      "epoch": 3.7824474660074165,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 0.8615,
+      "step": 15300
+    },
+    {
+      "epoch": 3.788627935723115,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.3974358974358978e-05,
+      "loss": 0.8603,
+      "step": 15325
+    },
+    {
+      "epoch": 3.7948084054388134,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.384615384615385e-05,
+      "loss": 0.8675,
+      "step": 15350
+    },
+    {
+      "epoch": 3.800988875154512,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.3717948717948718e-05,
+      "loss": 0.8655,
+      "step": 15375
+    },
+    {
+      "epoch": 3.80716934487021,
+      "grad_norm": 0.71875,
+      "learning_rate": 2.358974358974359e-05,
+      "loss": 0.8632,
+      "step": 15400
+    },
+    {
+      "epoch": 3.8133498145859086,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.3461538461538464e-05,
+      "loss": 0.8599,
+      "step": 15425
+    },
+    {
+      "epoch": 3.819530284301607,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.8656,
+      "step": 15450
+    },
+    {
+      "epoch": 3.8257107540173054,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.3205128205128207e-05,
+      "loss": 0.8638,
+      "step": 15475
+    },
+    {
+      "epoch": 3.831891223733004,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.8621,
+      "step": 15500
+    },
+    {
+      "epoch": 3.8380716934487022,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.2948717948717947e-05,
+      "loss": 0.8651,
+      "step": 15525
+    },
+    {
+      "epoch": 3.8442521631644007,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.2820512820512822e-05,
+      "loss": 0.8607,
+      "step": 15550
+    },
+    {
+      "epoch": 3.850432632880099,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.2692307692307694e-05,
+      "loss": 0.8632,
+      "step": 15575
+    },
+    {
+      "epoch": 3.856613102595797,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.2564102564102566e-05,
+      "loss": 0.8632,
+      "step": 15600
+    },
+    {
+      "epoch": 3.862793572311496,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.2435897435897437e-05,
+      "loss": 0.8631,
+      "step": 15625
+    },
+    {
+      "epoch": 3.868974042027194,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.230769230769231e-05,
+      "loss": 0.8594,
+      "step": 15650
+    },
+    {
+      "epoch": 3.8751545117428927,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.217948717948718e-05,
+      "loss": 0.8617,
+      "step": 15675
+    },
+    {
+      "epoch": 3.8813349814585907,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.2051282051282052e-05,
+      "loss": 0.8617,
+      "step": 15700
+    },
+    {
+      "epoch": 3.887515451174289,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.1923076923076924e-05,
+      "loss": 0.8629,
+      "step": 15725
+    },
+    {
+      "epoch": 3.8936959208899875,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 0.8639,
+      "step": 15750
+    },
+    {
+      "epoch": 3.899876390605686,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.8616,
+      "step": 15775
+    },
+    {
+      "epoch": 3.9060568603213843,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.1538461538461542e-05,
+      "loss": 0.865,
+      "step": 15800
+    },
+    {
+      "epoch": 3.9122373300370827,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.141025641025641e-05,
+      "loss": 0.8609,
+      "step": 15825
+    },
+    {
+      "epoch": 3.918417799752781,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.1282051282051282e-05,
+      "loss": 0.861,
+      "step": 15850
+    },
+    {
+      "epoch": 3.9245982694684796,
+      "grad_norm": 0.7109375,
+      "learning_rate": 2.1153846153846154e-05,
+      "loss": 0.8613,
+      "step": 15875
+    },
+    {
+      "epoch": 3.930778739184178,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.102564102564103e-05,
+      "loss": 0.8624,
+      "step": 15900
+    },
+    {
+      "epoch": 3.9369592088998764,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.08974358974359e-05,
+      "loss": 0.8633,
+      "step": 15925
+    },
+    {
+      "epoch": 3.943139678615575,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.0769230769230772e-05,
+      "loss": 0.8589,
+      "step": 15950
+    },
+    {
+      "epoch": 3.949320148331273,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.064102564102564e-05,
+      "loss": 0.8605,
+      "step": 15975
+    },
+    {
+      "epoch": 3.9555006180469716,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.8612,
+      "step": 16000
+    },
+    {
+      "epoch": 3.96168108776267,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.0384615384615387e-05,
+      "loss": 0.8612,
+      "step": 16025
+    },
+    {
+      "epoch": 3.9678615574783684,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.025641025641026e-05,
+      "loss": 0.861,
+      "step": 16050
+    },
+    {
+      "epoch": 3.974042027194067,
+      "grad_norm": 0.73046875,
+      "learning_rate": 2.012820512820513e-05,
+      "loss": 0.8593,
+      "step": 16075
+    },
+    {
+      "epoch": 3.9802224969097653,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2e-05,
+      "loss": 0.8612,
+      "step": 16100
+    },
+    {
+      "epoch": 3.9864029666254637,
+      "grad_norm": 0.71484375,
+      "learning_rate": 1.987179487179487e-05,
+      "loss": 0.8664,
+      "step": 16125
+    },
+    {
+      "epoch": 3.992583436341162,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 0.8599,
+      "step": 16150
+    },
+    {
+      "epoch": 3.99876390605686,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.8649,
+      "step": 16175
+    },
+    {
+      "epoch": 4.004944375772559,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.8264,
+      "step": 16200
+    },
+    {
+      "epoch": 4.011124845488257,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 0.8184,
+      "step": 16225
+    },
+    {
+      "epoch": 4.017305315203956,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.8214,
+      "step": 16250
+    },
+    {
+      "epoch": 4.023485784919654,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.9102564102564103e-05,
+      "loss": 0.8166,
+      "step": 16275
+    },
+    {
+      "epoch": 4.0296662546353526,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.8191,
+      "step": 16300
+    },
+    {
+      "epoch": 4.0358467243510505,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.8194,
+      "step": 16325
+    },
+    {
+      "epoch": 4.042027194066749,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.8189,
+      "step": 16350
+    },
+    {
+      "epoch": 4.048207663782447,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.858974358974359e-05,
+      "loss": 0.822,
+      "step": 16375
+    },
+    {
+      "epoch": 4.054388133498146,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.8146,
+      "step": 16400
+    },
+    {
+      "epoch": 4.060568603213844,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.8168,
+      "step": 16425
+    },
+    {
+      "epoch": 4.066749072929543,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.8205128205128204e-05,
+      "loss": 0.8221,
+      "step": 16450
+    },
+    {
+      "epoch": 4.072929542645241,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.8076923076923076e-05,
+      "loss": 0.8179,
+      "step": 16475
+    },
+    {
+      "epoch": 4.07911001236094,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.8182,
+      "step": 16500
+    },
+    {
+      "epoch": 4.085290482076638,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 0.8145,
+      "step": 16525
+    },
+    {
+      "epoch": 4.091470951792337,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.8239,
+      "step": 16550
+    },
+    {
+      "epoch": 4.097651421508035,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.7564102564102563e-05,
+      "loss": 0.8151,
+      "step": 16575
+    },
+    {
+      "epoch": 4.103831891223733,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.7435897435897434e-05,
+      "loss": 0.8193,
+      "step": 16600
+    },
+    {
+      "epoch": 4.1100123609394315,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.8193,
+      "step": 16625
+    },
+    {
+      "epoch": 4.116192830655129,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 0.8193,
+      "step": 16650
+    },
+    {
+      "epoch": 4.122373300370828,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 0.8214,
+      "step": 16675
+    },
+    {
+      "epoch": 4.128553770086526,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.8211,
+      "step": 16700
+    },
+    {
+      "epoch": 4.134734239802225,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 0.8213,
+      "step": 16725
+    },
+    {
+      "epoch": 4.140914709517923,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.8209,
+      "step": 16750
+    },
+    {
+      "epoch": 4.147095179233622,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.8185,
+      "step": 16775
+    },
+    {
+      "epoch": 4.15327564894932,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.8188,
+      "step": 16800
+    },
+    {
+      "epoch": 4.159456118665019,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 0.8209,
+      "step": 16825
+    },
+    {
+      "epoch": 4.165636588380717,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.82,
+      "step": 16850
+    },
+    {
+      "epoch": 4.171817058096416,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.8203,
+      "step": 16875
+    },
+    {
+      "epoch": 4.1779975278121135,
+      "grad_norm": 0.71484375,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.8175,
+      "step": 16900
+    },
+    {
+      "epoch": 4.184177997527812,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.8207,
+      "step": 16925
+    },
+    {
+      "epoch": 4.19035846724351,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.564102564102564e-05,
+      "loss": 0.8213,
+      "step": 16950
+    },
+    {
+      "epoch": 4.196538936959209,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 0.8235,
+      "step": 16975
+    },
+    {
+      "epoch": 4.202719406674907,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.8166,
+      "step": 17000
+    },
+    {
+      "epoch": 4.208899876390606,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.5256410256410259e-05,
+      "loss": 0.8178,
+      "step": 17025
+    },
+    {
+      "epoch": 4.215080346106304,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.8214,
+      "step": 17050
+    },
+    {
+      "epoch": 4.221260815822003,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8228,
+      "step": 17075
+    },
+    {
+      "epoch": 4.227441285537701,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.4871794871794872e-05,
+      "loss": 0.8216,
+      "step": 17100
+    },
+    {
+      "epoch": 4.2336217552534,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.8202,
+      "step": 17125
+    },
+    {
+      "epoch": 4.239802224969098,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.4615384615384617e-05,
+      "loss": 0.8245,
+      "step": 17150
+    },
+    {
+      "epoch": 4.2459826946847965,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 0.8228,
+      "step": 17175
+    },
+    {
+      "epoch": 4.2521631644004945,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.4358974358974359e-05,
+      "loss": 0.8222,
+      "step": 17200
+    },
+    {
+      "epoch": 4.258343634116192,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.423076923076923e-05,
+      "loss": 0.8213,
+      "step": 17225
+    },
+    {
+      "epoch": 4.264524103831891,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 0.8226,
+      "step": 17250
+    },
+    {
+      "epoch": 4.270704573547589,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 0.8171,
+      "step": 17275
+    },
+    {
+      "epoch": 4.276885043263288,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.8161,
+      "step": 17300
+    },
+    {
+      "epoch": 4.283065512978986,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.371794871794872e-05,
+      "loss": 0.8203,
+      "step": 17325
+    },
+    {
+      "epoch": 4.289245982694685,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.358974358974359e-05,
+      "loss": 0.8184,
+      "step": 17350
+    },
+    {
+      "epoch": 4.295426452410383,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.3461538461538462e-05,
+      "loss": 0.8223,
+      "step": 17375
+    },
+    {
+      "epoch": 4.301606922126082,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8207,
+      "step": 17400
+    },
+    {
+      "epoch": 4.30778739184178,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 0.8233,
+      "step": 17425
+    },
+    {
+      "epoch": 4.313967861557479,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.8211,
+      "step": 17450
+    },
+    {
+      "epoch": 4.3201483312731765,
+      "grad_norm": 0.73046875,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 0.8193,
+      "step": 17475
+    },
+    {
+      "epoch": 4.326328800988875,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 0.8207,
+      "step": 17500
+    },
+    {
+      "epoch": 4.332509270704573,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.2692307692307691e-05,
+      "loss": 0.8247,
+      "step": 17525
+    },
+    {
+      "epoch": 4.338689740420272,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.826,
+      "step": 17550
+    },
+    {
+      "epoch": 4.34487021013597,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 0.82,
+      "step": 17575
+    },
+    {
+      "epoch": 4.351050679851669,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.8161,
+      "step": 17600
+    },
+    {
+      "epoch": 4.357231149567367,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.826,
+      "step": 17625
+    },
+    {
+      "epoch": 4.363411619283066,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.8199,
+      "step": 17650
+    },
+    {
+      "epoch": 4.369592088998764,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.8251,
+      "step": 17675
+    },
+    {
+      "epoch": 4.375772558714463,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1794871794871795e-05,
+      "loss": 0.8209,
+      "step": 17700
+    },
+    {
+      "epoch": 4.381953028430161,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.8206,
+      "step": 17725
+    },
+    {
+      "epoch": 4.388133498145859,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 0.8209,
+      "step": 17750
+    },
+    {
+      "epoch": 4.3943139678615575,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 0.8218,
+      "step": 17775
+    },
+    {
+      "epoch": 4.400494437577256,
+      "grad_norm": 0.68359375,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.8216,
+      "step": 17800
+    },
+    {
+      "epoch": 4.406674907292954,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.82,
+      "step": 17825
+    },
+    {
+      "epoch": 4.412855377008652,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.1025641025641026e-05,
+      "loss": 0.8224,
+      "step": 17850
+    },
+    {
+      "epoch": 4.419035846724351,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.8209,
+      "step": 17875
+    },
+    {
+      "epoch": 4.425216316440049,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.0769230769230771e-05,
+      "loss": 0.8198,
+      "step": 17900
+    },
+    {
+      "epoch": 4.431396786155748,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0641025641025641e-05,
+      "loss": 0.8259,
+      "step": 17925
+    },
+    {
+      "epoch": 4.437577255871446,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.8211,
+      "step": 17950
+    },
+    {
+      "epoch": 4.443757725587145,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.8213,
+      "step": 17975
+    },
+    {
+      "epoch": 4.449938195302843,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.8266,
+      "step": 18000
+    },
+    {
+      "epoch": 4.456118665018542,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 0.821,
+      "step": 18025
+    },
+    {
+      "epoch": 4.46229913473424,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1e-05,
+      "loss": 0.8223,
+      "step": 18050
+    },
+    {
+      "epoch": 4.468479604449938,
+      "grad_norm": 0.6953125,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 0.8216,
+      "step": 18075
+    },
+    {
+      "epoch": 4.474660074165636,
+      "grad_norm": 0.703125,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.8183,
+      "step": 18100
+    },
+    {
+      "epoch": 4.480840543881335,
+      "grad_norm": 0.6953125,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.8243,
+      "step": 18125
+    },
+    {
+      "epoch": 4.487021013597033,
+      "grad_norm": 0.71484375,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.8197,
+      "step": 18150
+    },
+    {
+      "epoch": 4.493201483312732,
+      "grad_norm": 0.71484375,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 0.8243,
+      "step": 18175
+    },
+    {
+      "epoch": 4.49938195302843,
+      "grad_norm": 0.703125,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.8211,
+      "step": 18200
+    },
+    {
+      "epoch": 4.505562422744129,
+      "grad_norm": 0.6953125,
+      "learning_rate": 9.102564102564102e-06,
+      "loss": 0.8222,
+      "step": 18225
+    },
+    {
+      "epoch": 4.511742892459827,
+      "grad_norm": 0.6953125,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.8225,
+      "step": 18250
+    },
+    {
+      "epoch": 4.517923362175526,
+      "grad_norm": 0.69140625,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.8261,
+      "step": 18275
+    },
+    {
+      "epoch": 4.524103831891224,
+      "grad_norm": 0.6796875,
+      "learning_rate": 8.717948717948717e-06,
+      "loss": 0.8234,
+      "step": 18300
+    },
+    {
+      "epoch": 4.5302843016069225,
+      "grad_norm": 0.6796875,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 0.8213,
+      "step": 18325
+    },
+    {
+      "epoch": 4.5364647713226205,
+      "grad_norm": 0.70703125,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.8272,
+      "step": 18350
+    },
+    {
+      "epoch": 4.5426452410383185,
+      "grad_norm": 0.6796875,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.8223,
+      "step": 18375
+    },
+    {
+      "epoch": 4.548825710754017,
+      "grad_norm": 0.70703125,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.8218,
+      "step": 18400
+    },
+    {
+      "epoch": 4.555006180469716,
+      "grad_norm": 0.6875,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.8195,
+      "step": 18425
+    },
+    {
+      "epoch": 4.561186650185414,
+      "grad_norm": 0.6640625,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.8222,
+      "step": 18450
+    },
+    {
+      "epoch": 4.567367119901112,
+      "grad_norm": 0.69140625,
+      "learning_rate": 7.82051282051282e-06,
+      "loss": 0.8212,
+      "step": 18475
+    },
+    {
+      "epoch": 4.573547589616811,
+      "grad_norm": 0.703125,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.8263,
+      "step": 18500
+    },
+    {
+      "epoch": 4.579728059332509,
+      "grad_norm": 0.67578125,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.824,
+      "step": 18525
+    },
+    {
+      "epoch": 4.585908529048208,
+      "grad_norm": 0.7109375,
+      "learning_rate": 7.435897435897436e-06,
+      "loss": 0.8189,
+      "step": 18550
+    },
+    {
+      "epoch": 4.592088998763906,
+      "grad_norm": 0.6953125,
+      "learning_rate": 7.3076923076923085e-06,
+      "loss": 0.8213,
+      "step": 18575
+    },
+    {
+      "epoch": 4.598269468479605,
+      "grad_norm": 0.6796875,
+      "learning_rate": 7.179487179487179e-06,
+      "loss": 0.8183,
+      "step": 18600
+    },
+    {
+      "epoch": 4.604449938195303,
+      "grad_norm": 0.6875,
+      "learning_rate": 7.051282051282052e-06,
+      "loss": 0.8203,
+      "step": 18625
+    },
+    {
+      "epoch": 4.610630407911001,
+      "grad_norm": 0.671875,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.819,
+      "step": 18650
+    },
+    {
+      "epoch": 4.616810877626699,
+      "grad_norm": 0.73828125,
+      "learning_rate": 6.794871794871795e-06,
+      "loss": 0.8235,
+      "step": 18675
+    },
+    {
+      "epoch": 4.622991347342398,
+      "grad_norm": 0.69140625,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8229,
+      "step": 18700
+    },
+    {
+      "epoch": 4.629171817058096,
+      "grad_norm": 0.6796875,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.8189,
+      "step": 18725
+    },
+    {
+      "epoch": 4.635352286773795,
+      "grad_norm": 0.6875,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 0.8224,
+      "step": 18750
+    },
+    {
+      "epoch": 4.641532756489493,
+      "grad_norm": 0.66796875,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 0.8225,
+      "step": 18775
+    },
+    {
+      "epoch": 4.647713226205192,
+      "grad_norm": 0.67578125,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.8194,
+      "step": 18800
+    },
+    {
+      "epoch": 4.65389369592089,
+      "grad_norm": 0.68359375,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 0.8225,
+      "step": 18825
+    },
+    {
+      "epoch": 4.660074165636589,
+      "grad_norm": 0.6875,
+      "learning_rate": 5.897435897435897e-06,
+      "loss": 0.8169,
+      "step": 18850
+    },
+    {
+      "epoch": 4.666254635352287,
+      "grad_norm": 0.6953125,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 0.8153,
+      "step": 18875
+    },
+    {
+      "epoch": 4.6724351050679855,
+      "grad_norm": 0.671875,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.822,
+      "step": 18900
+    },
+    {
+      "epoch": 4.6786155747836835,
+      "grad_norm": 0.671875,
+      "learning_rate": 5.512820512820513e-06,
+      "loss": 0.822,
+      "step": 18925
+    },
+    {
+      "epoch": 4.684796044499382,
+      "grad_norm": 0.6796875,
+      "learning_rate": 5.3846153846153855e-06,
+      "loss": 0.8184,
+      "step": 18950
+    },
+    {
+      "epoch": 4.69097651421508,
+      "grad_norm": 0.69921875,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 0.8247,
+      "step": 18975
+    },
+    {
+      "epoch": 4.697156983930778,
+      "grad_norm": 0.67578125,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.8212,
+      "step": 19000
+    },
+    {
+      "epoch": 4.703337453646477,
+      "grad_norm": 0.671875,
+      "learning_rate": 5e-06,
+      "loss": 0.8226,
+      "step": 19025
+    },
+    {
+      "epoch": 4.709517923362176,
+      "grad_norm": 0.66796875,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 0.8206,
+      "step": 19050
+    },
+    {
+      "epoch": 4.715698393077874,
+      "grad_norm": 0.6796875,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 0.8183,
+      "step": 19075
+    },
+    {
+      "epoch": 4.721878862793572,
+      "grad_norm": 0.69140625,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.8184,
+      "step": 19100
+    },
+    {
+      "epoch": 4.728059332509271,
+      "grad_norm": 0.70703125,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 0.8212,
+      "step": 19125
+    },
+    {
+      "epoch": 4.734239802224969,
+      "grad_norm": 0.66796875,
+      "learning_rate": 4.3589743589743586e-06,
+      "loss": 0.8198,
+      "step": 19150
+    },
+    {
+      "epoch": 4.740420271940668,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 0.8235,
+      "step": 19175
+    },
+    {
+      "epoch": 4.746600741656366,
+      "grad_norm": 0.6796875,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.8228,
+      "step": 19200
+    },
+    {
+      "epoch": 4.752781211372064,
+      "grad_norm": 0.67578125,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 0.8278,
+      "step": 19225
+    },
+    {
+      "epoch": 4.758961681087762,
+      "grad_norm": 0.66015625,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.817,
+      "step": 19250
+    },
+    {
+      "epoch": 4.765142150803461,
+      "grad_norm": 0.6953125,
+      "learning_rate": 3.717948717948718e-06,
+      "loss": 0.8226,
+      "step": 19275
+    },
+    {
+      "epoch": 4.771322620519159,
+      "grad_norm": 0.6796875,
+      "learning_rate": 3.5897435897435896e-06,
+      "loss": 0.8218,
+      "step": 19300
+    },
+    {
+      "epoch": 4.777503090234858,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.8209,
+      "step": 19325
+    },
+    {
+      "epoch": 4.783683559950556,
+      "grad_norm": 0.6875,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8231,
+      "step": 19350
+    },
+    {
+      "epoch": 4.789864029666255,
+      "grad_norm": 0.6875,
+      "learning_rate": 3.205128205128205e-06,
+      "loss": 0.8199,
+      "step": 19375
+    },
+    {
+      "epoch": 4.796044499381953,
+      "grad_norm": 0.68359375,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.8214,
+      "step": 19400
+    },
+    {
+      "epoch": 4.802224969097652,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.9487179487179486e-06,
+      "loss": 0.8221,
+      "step": 19425
+    },
+    {
+      "epoch": 4.80840543881335,
+      "grad_norm": 0.6796875,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 0.8198,
+      "step": 19450
+    },
+    {
+      "epoch": 4.8145859085290486,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.6923076923076928e-06,
+      "loss": 0.8195,
+      "step": 19475
+    },
+    {
+      "epoch": 4.8207663782447465,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.8217,
+      "step": 19500
+    },
+    {
+      "epoch": 4.826946847960445,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 0.8185,
+      "step": 19525
+    },
+    {
+      "epoch": 4.833127317676143,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.8196,
+      "step": 19550
+    },
+    {
+      "epoch": 4.839307787391842,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.1794871794871793e-06,
+      "loss": 0.8204,
+      "step": 19575
+    },
+    {
+      "epoch": 4.84548825710754,
+      "grad_norm": 0.66796875,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 0.8201,
+      "step": 19600
+    },
+    {
+      "epoch": 4.851668726823238,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.8229,
+      "step": 19625
+    },
+    {
+      "epoch": 4.857849196538937,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.7948717948717948e-06,
+      "loss": 0.8222,
+      "step": 19650
+    },
+    {
+      "epoch": 4.864029666254636,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.8232,
+      "step": 19675
+    },
+    {
+      "epoch": 4.870210135970334,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.8224,
+      "step": 19700
+    },
+    {
+      "epoch": 4.876390605686032,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 0.8235,
+      "step": 19725
+    },
+    {
+      "epoch": 4.882571075401731,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.8164,
+      "step": 19750
+    },
+    {
+      "epoch": 4.888751545117429,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 0.8217,
+      "step": 19775
+    },
+    {
+      "epoch": 4.8949320148331275,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 0.8185,
+      "step": 19800
+    },
+    {
+      "epoch": 4.901112484548825,
+      "grad_norm": 0.66796875,
+      "learning_rate": 8.974358974358974e-07,
+      "loss": 0.8256,
+      "step": 19825
+    },
+    {
+      "epoch": 4.907292954264524,
+      "grad_norm": 0.68359375,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 0.8235,
+      "step": 19850
+    },
+    {
+      "epoch": 4.913473423980222,
+      "grad_norm": 0.67578125,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 0.8234,
+      "step": 19875
+    },
+    {
+      "epoch": 4.919653893695921,
+      "grad_norm": 0.6953125,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 0.8208,
+      "step": 19900
+    },
+    {
+      "epoch": 4.925834363411619,
+      "grad_norm": 0.68359375,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 0.8235,
+      "step": 19925
+    },
+    {
+      "epoch": 4.932014833127318,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 0.8218,
+      "step": 19950
+    },
+    {
+      "epoch": 4.938195302843016,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 0.8207,
+      "step": 19975
+    },
+    {
+      "epoch": 4.944375772558715,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0,
+      "loss": 0.8191,
+      "step": 20000
+    },
+    {
+      "epoch": 4.944375772558715,
+      "eval_loss": 1.0042184591293335,
+      "eval_runtime": 1.5308,
+      "eval_samples_per_second": 417.441,
+      "eval_steps_per_second": 1.96,
+      "step": 20000
+    },
+    {
+      "epoch": 4.944375772558715,
+      "step": 20000,
+      "total_flos": 9.058112140235663e+19,
+      "train_loss": 1.0958750234603882,
+      "train_runtime": 36068.0493,
+      "train_samples_per_second": 141.954,
+      "train_steps_per_second": 0.555
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 20000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 5000,
+  "total_flos": 9.058112140235663e+19,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}