diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2337 +1,2915 @@
 {
-  "best_metric": 13.816958025614658,
-  "best_model_checkpoint": "./checkpoint-8000",
+  "best_metric": 12.307980517047582,
+  "best_model_checkpoint": "./checkpoint-10000",
   "epoch": 1.0,
   "eval_steps": 1000,
-  "global_step": 8000,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.003125,
-      "grad_norm": 21.98771095275879,
-      "learning_rate": 8.8e-07,
-      "loss": 3.1382,
+      "epoch": 0.0025,
+      "grad_norm": 20.66425132751465,
+      "learning_rate": 1.1e-06,
+      "loss": 3.1217,
       "step": 25
     },
     {
-      "epoch": 0.00625,
-      "grad_norm": 11.417325019836426,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 2.4524,
+      "epoch": 0.005,
+      "grad_norm": 10.962828636169434,
+      "learning_rate": 2.35e-06,
+      "loss": 2.3804,
       "step": 50
     },
     {
-      "epoch": 0.009375,
-      "grad_norm": 9.939048767089844,
-      "learning_rate": 2.88e-06,
-      "loss": 1.6386,
+      "epoch": 0.0075,
+      "grad_norm": 9.967100143432617,
+      "learning_rate": 3.6e-06,
+      "loss": 1.546,
       "step": 75
     },
     {
-      "epoch": 0.0125,
-      "grad_norm": 8.167191505432129,
-      "learning_rate": 3.88e-06,
-      "loss": 1.26,
+      "epoch": 0.01,
+      "grad_norm": 7.833068370819092,
+      "learning_rate": 4.85e-06,
+      "loss": 1.1938,
       "step": 100
     },
     {
-      "epoch": 0.015625,
-      "grad_norm": 7.114121913909912,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 1.0564,
+      "epoch": 0.0125,
+      "grad_norm": 6.779320240020752,
+      "learning_rate": 6.1e-06,
+      "loss": 0.9869,
       "step": 125
     },
     {
-      "epoch": 0.01875,
-      "grad_norm": 6.8056840896606445,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 0.897,
+      "epoch": 0.015,
+      "grad_norm": 6.599488735198975,
+      "learning_rate": 7.35e-06,
+      "loss": 0.8336,
       "step": 150
     },
     {
-      "epoch": 0.021875,
-      "grad_norm": 6.363353252410889,
-      "learning_rate": 6.88e-06,
-      "loss": 0.8252,
+      "epoch": 0.0175,
+      "grad_norm": 6.097827911376953,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 0.7657,
       "step": 175
     },
     {
-      "epoch": 0.025,
-      "grad_norm": 6.622057914733887,
-      "learning_rate": 7.88e-06,
-      "loss": 0.7693,
+      "epoch": 0.02,
+      "grad_norm": 6.401536464691162,
+      "learning_rate": 9.85e-06,
+      "loss": 0.7115,
       "step": 200
     },
     {
-      "epoch": 0.028125,
-      "grad_norm": 5.045984745025635,
-      "learning_rate": 8.880000000000001e-06,
-      "loss": 0.6621,
+      "epoch": 0.0225,
+      "grad_norm": 4.918787479400635,
+      "learning_rate": 1.11e-05,
+      "loss": 0.6108,
       "step": 225
     },
     {
-      "epoch": 0.03125,
-      "grad_norm": 7.978261947631836,
-      "learning_rate": 9.88e-06,
-      "loss": 0.6861,
+      "epoch": 0.025,
+      "grad_norm": 7.6149492263793945,
+      "learning_rate": 1.235e-05,
+      "loss": 0.6382,
       "step": 250
     },
     {
-      "epoch": 0.034375,
-      "grad_norm": 6.535711288452148,
-      "learning_rate": 1.0880000000000001e-05,
-      "loss": 0.6888,
+      "epoch": 0.0275,
+      "grad_norm": 6.412052631378174,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.6505,
       "step": 275
     },
     {
-      "epoch": 0.0375,
-      "grad_norm": 6.781430721282959,
-      "learning_rate": 1.188e-05,
-      "loss": 0.6648,
+      "epoch": 0.03,
+      "grad_norm": 6.510907173156738,
+      "learning_rate": 1.485e-05,
+      "loss": 0.6308,
       "step": 300
     },
     {
-      "epoch": 0.040625,
-      "grad_norm": 5.826904773712158,
-      "learning_rate": 1.2880000000000002e-05,
-      "loss": 0.5983,
+      "epoch": 0.0325,
+      "grad_norm": 5.585999965667725,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.5634,
       "step": 325
     },
     {
-      "epoch": 0.04375,
-      "grad_norm": 5.830564975738525,
-      "learning_rate": 1.3880000000000001e-05,
-      "loss": 0.5272,
+      "epoch": 0.035,
+      "grad_norm": 5.468475341796875,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 0.488,
       "step": 350
     },
     {
-      "epoch": 0.046875,
-      "grad_norm": 5.638543128967285,
-      "learning_rate": 1.4880000000000002e-05,
-      "loss": 0.4479,
+      "epoch": 0.0375,
+      "grad_norm": 5.375606536865234,
+      "learning_rate": 1.86e-05,
+      "loss": 0.4121,
       "step": 375
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 4.451868057250977,
-      "learning_rate": 1.588e-05,
-      "loss": 0.4341,
+      "epoch": 0.04,
+      "grad_norm": 4.386756896972656,
+      "learning_rate": 1.985e-05,
+      "loss": 0.4017,
       "step": 400
     },
     {
-      "epoch": 0.053125,
-      "grad_norm": 4.475216865539551,
-      "learning_rate": 1.688e-05,
-      "loss": 0.3916,
+      "epoch": 0.0425,
+      "grad_norm": 4.322641372680664,
+      "learning_rate": 2.11e-05,
+      "loss": 0.3617,
       "step": 425
     },
     {
-      "epoch": 0.05625,
-      "grad_norm": 4.683038711547852,
-      "learning_rate": 1.788e-05,
-      "loss": 0.3745,
+      "epoch": 0.045,
+      "grad_norm": 4.434195041656494,
+      "learning_rate": 2.235e-05,
+      "loss": 0.3469,
       "step": 450
     },
     {
-      "epoch": 0.059375,
-      "grad_norm": 4.93367862701416,
-      "learning_rate": 1.8880000000000002e-05,
-      "loss": 0.3569,
+      "epoch": 0.0475,
+      "grad_norm": 4.9860920906066895,
+      "learning_rate": 2.36e-05,
+      "loss": 0.331,
       "step": 475
     },
     {
-      "epoch": 0.0625,
-      "grad_norm": 4.0592041015625,
-      "learning_rate": 1.9880000000000003e-05,
-      "loss": 0.3336,
+      "epoch": 0.05,
+      "grad_norm": 3.9041032791137695,
+      "learning_rate": 2.485e-05,
+      "loss": 0.3101,
       "step": 500
     },
     {
-      "epoch": 0.065625,
-      "grad_norm": 5.144535064697266,
-      "learning_rate": 1.9941333333333335e-05,
-      "loss": 0.3092,
+      "epoch": 0.0525,
+      "grad_norm": 5.439699649810791,
+      "learning_rate": 2.4942105263157897e-05,
+      "loss": 0.2881,
       "step": 525
     },
     {
-      "epoch": 0.06875,
-      "grad_norm": 4.806638717651367,
-      "learning_rate": 1.9874666666666668e-05,
-      "loss": 0.3106,
+      "epoch": 0.055,
+      "grad_norm": 3.6921448707580566,
+      "learning_rate": 2.4876315789473686e-05,
+      "loss": 0.2906,
       "step": 550
     },
     {
-      "epoch": 0.071875,
-      "grad_norm": 4.3809638023376465,
-      "learning_rate": 1.9808e-05,
-      "loss": 0.2971,
+      "epoch": 0.0575,
+      "grad_norm": 4.33128547668457,
+      "learning_rate": 2.4810526315789475e-05,
+      "loss": 0.2762,
       "step": 575
     },
     {
-      "epoch": 0.075,
-      "grad_norm": 5.35611629486084,
-      "learning_rate": 1.9741333333333334e-05,
-      "loss": 0.2897,
+      "epoch": 0.06,
+      "grad_norm": 5.252212047576904,
+      "learning_rate": 2.4744736842105264e-05,
+      "loss": 0.2692,
       "step": 600
     },
     {
-      "epoch": 0.078125,
-      "grad_norm": 4.62730598449707,
-      "learning_rate": 1.967466666666667e-05,
-      "loss": 0.2607,
+      "epoch": 0.0625,
+      "grad_norm": 4.352893829345703,
+      "learning_rate": 2.4678947368421053e-05,
+      "loss": 0.2429,
       "step": 625
     },
     {
-      "epoch": 0.08125,
-      "grad_norm": 3.60905122756958,
-      "learning_rate": 1.9608000000000003e-05,
-      "loss": 0.2765,
+      "epoch": 0.065,
+      "grad_norm": 3.3367650508880615,
+      "learning_rate": 2.4613157894736842e-05,
+      "loss": 0.2587,
       "step": 650
     },
     {
-      "epoch": 0.084375,
-      "grad_norm": 3.7476887702941895,
-      "learning_rate": 1.9541333333333336e-05,
-      "loss": 0.2661,
+      "epoch": 0.0675,
+      "grad_norm": 3.517099618911743,
+      "learning_rate": 2.454736842105263e-05,
+      "loss": 0.2484,
       "step": 675
     },
     {
-      "epoch": 0.0875,
-      "grad_norm": 4.03351354598999,
-      "learning_rate": 1.947466666666667e-05,
-      "loss": 0.271,
+      "epoch": 0.07,
+      "grad_norm": 4.486610412597656,
+      "learning_rate": 2.448157894736842e-05,
+      "loss": 0.2544,
       "step": 700
     },
     {
-      "epoch": 0.090625,
-      "grad_norm": 6.19671106338501,
-      "learning_rate": 1.9408e-05,
-      "loss": 0.3045,
+      "epoch": 0.0725,
+      "grad_norm": 6.344348430633545,
+      "learning_rate": 2.441578947368421e-05,
+      "loss": 0.2895,
       "step": 725
     },
     {
-      "epoch": 0.09375,
-      "grad_norm": 6.079224586486816,
-      "learning_rate": 1.9341333333333334e-05,
-      "loss": 0.4168,
+      "epoch": 0.075,
+      "grad_norm": 5.6651530265808105,
+      "learning_rate": 2.435e-05,
+      "loss": 0.4015,
       "step": 750
     },
     {
-      "epoch": 0.096875,
-      "grad_norm": 5.619544982910156,
-      "learning_rate": 1.9274666666666667e-05,
-      "loss": 0.4038,
+      "epoch": 0.0775,
+      "grad_norm": 5.023654937744141,
+      "learning_rate": 2.428421052631579e-05,
+      "loss": 0.3926,
       "step": 775
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 5.917449951171875,
-      "learning_rate": 1.9208000000000003e-05,
-      "loss": 0.4418,
+      "epoch": 0.08,
+      "grad_norm": 5.747288703918457,
+      "learning_rate": 2.421842105263158e-05,
+      "loss": 0.4282,
       "step": 800
     },
     {
-      "epoch": 0.103125,
-      "grad_norm": 3.9295430183410645,
-      "learning_rate": 1.9141333333333333e-05,
-      "loss": 0.3051,
+      "epoch": 0.0825,
+      "grad_norm": 3.902658700942993,
+      "learning_rate": 2.415263157894737e-05,
+      "loss": 0.2929,
       "step": 825
     },
     {
-      "epoch": 0.10625,
-      "grad_norm": 3.26605486869812,
-      "learning_rate": 1.907466666666667e-05,
-      "loss": 0.2447,
+      "epoch": 0.085,
+      "grad_norm": 3.1161112785339355,
+      "learning_rate": 2.408684210526316e-05,
+      "loss": 0.2316,
       "step": 850
     },
     {
-      "epoch": 0.109375,
-      "grad_norm": 4.2773051261901855,
-      "learning_rate": 1.9008e-05,
-      "loss": 0.2316,
+      "epoch": 0.0875,
+      "grad_norm": 3.6460275650024414,
+      "learning_rate": 2.402105263157895e-05,
+      "loss": 0.2167,
       "step": 875
     },
     {
-      "epoch": 0.1125,
-      "grad_norm": 5.690479755401611,
-      "learning_rate": 1.8941333333333334e-05,
-      "loss": 0.3193,
+      "epoch": 0.09,
+      "grad_norm": 5.112854480743408,
+      "learning_rate": 2.395526315789474e-05,
+      "loss": 0.3073,
       "step": 900
     },
     {
-      "epoch": 0.115625,
-      "grad_norm": 5.861849308013916,
-      "learning_rate": 1.8874666666666667e-05,
-      "loss": 0.3742,
+      "epoch": 0.0925,
+      "grad_norm": 5.572996139526367,
+      "learning_rate": 2.388947368421053e-05,
+      "loss": 0.3637,
       "step": 925
     },
     {
-      "epoch": 0.11875,
-      "grad_norm": 5.605452537536621,
-      "learning_rate": 1.8808e-05,
-      "loss": 0.3622,
+      "epoch": 0.095,
+      "grad_norm": 5.3848490715026855,
+      "learning_rate": 2.382368421052632e-05,
+      "loss": 0.3493,
       "step": 950
     },
     {
-      "epoch": 0.121875,
-      "grad_norm": 6.191706657409668,
-      "learning_rate": 1.8741333333333336e-05,
-      "loss": 0.5693,
+      "epoch": 0.0975,
+      "grad_norm": 6.014389991760254,
+      "learning_rate": 2.3757894736842108e-05,
+      "loss": 0.5539,
       "step": 975
     },
     {
-      "epoch": 0.125,
-      "grad_norm": 6.441635608673096,
-      "learning_rate": 1.867466666666667e-05,
-      "loss": 0.4951,
+      "epoch": 0.1,
+      "grad_norm": 6.122930526733398,
+      "learning_rate": 2.3692105263157897e-05,
+      "loss": 0.4816,
       "step": 1000
     },
     {
-      "epoch": 0.125,
-      "eval_loss": 0.49006596207618713,
-      "eval_runtime": 77.9711,
-      "eval_samples_per_second": 26.984,
-      "eval_steps_per_second": 1.693,
-      "eval_wer": 27.05431429372721,
+      "epoch": 0.1,
+      "eval_loss": 0.513611376285553,
+      "eval_runtime": 50.853,
+      "eval_samples_per_second": 33.253,
+      "eval_steps_per_second": 2.084,
+      "eval_wer": 25.752466591732237,
       "step": 1000
     },
     {
-      "epoch": 0.128125,
-      "grad_norm": 5.985304832458496,
-      "learning_rate": 1.8608000000000002e-05,
-      "loss": 0.4113,
+      "epoch": 0.1025,
+      "grad_norm": 5.829473495483398,
+      "learning_rate": 2.3626315789473686e-05,
+      "loss": 0.3987,
       "step": 1025
     },
     {
-      "epoch": 0.13125,
-      "grad_norm": 5.439868450164795,
-      "learning_rate": 1.8541333333333335e-05,
-      "loss": 0.3116,
+      "epoch": 0.105,
+      "grad_norm": 4.1634039878845215,
+      "learning_rate": 2.3560526315789475e-05,
+      "loss": 0.3008,
       "step": 1050
     },
     {
-      "epoch": 0.134375,
-      "grad_norm": 3.6065237522125244,
-      "learning_rate": 1.8474666666666668e-05,
-      "loss": 0.223,
+      "epoch": 0.1075,
+      "grad_norm": 3.4256279468536377,
+      "learning_rate": 2.3494736842105264e-05,
+      "loss": 0.2114,
       "step": 1075
     },
     {
-      "epoch": 0.1375,
-      "grad_norm": 3.4091665744781494,
-      "learning_rate": 1.8408e-05,
-      "loss": 0.1918,
+      "epoch": 0.11,
+      "grad_norm": 3.1974499225616455,
+      "learning_rate": 2.3428947368421053e-05,
+      "loss": 0.1819,
       "step": 1100
     },
     {
-      "epoch": 0.140625,
-      "grad_norm": 2.80391001701355,
-      "learning_rate": 1.8341333333333337e-05,
-      "loss": 0.1784,
+      "epoch": 0.1125,
+      "grad_norm": 3.0676839351654053,
+      "learning_rate": 2.3363157894736843e-05,
+      "loss": 0.1678,
       "step": 1125
     },
     {
-      "epoch": 0.14375,
-      "grad_norm": 3.194566011428833,
-      "learning_rate": 1.8274666666666666e-05,
-      "loss": 0.1988,
+      "epoch": 0.115,
+      "grad_norm": 3.030893087387085,
+      "learning_rate": 2.329736842105263e-05,
+      "loss": 0.1886,
       "step": 1150
     },
     {
-      "epoch": 0.146875,
-      "grad_norm": 3.294611930847168,
-      "learning_rate": 1.8208000000000003e-05,
-      "loss": 0.2032,
+      "epoch": 0.1175,
+      "grad_norm": 3.1832079887390137,
+      "learning_rate": 2.323157894736842e-05,
+      "loss": 0.1941,
       "step": 1175
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 3.653074264526367,
-      "learning_rate": 1.8141333333333335e-05,
-      "loss": 0.1914,
+      "epoch": 0.12,
+      "grad_norm": 4.08896541595459,
+      "learning_rate": 2.3165789473684213e-05,
+      "loss": 0.1822,
       "step": 1200
     },
     {
-      "epoch": 0.153125,
-      "grad_norm": 4.978360176086426,
-      "learning_rate": 1.8074666666666668e-05,
-      "loss": 0.2865,
+      "epoch": 0.1225,
+      "grad_norm": 4.830000400543213,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 0.2749,
       "step": 1225
     },
     {
-      "epoch": 0.15625,
-      "grad_norm": 5.2725701332092285,
-      "learning_rate": 1.8008e-05,
-      "loss": 0.306,
+      "epoch": 0.125,
+      "grad_norm": 4.692865371704102,
+      "learning_rate": 2.303421052631579e-05,
+      "loss": 0.2947,
       "step": 1250
     },
     {
-      "epoch": 0.159375,
-      "grad_norm": 5.122636318206787,
-      "learning_rate": 1.7941333333333334e-05,
-      "loss": 0.3283,
+      "epoch": 0.1275,
+      "grad_norm": 4.8690571784973145,
+      "learning_rate": 2.296842105263158e-05,
+      "loss": 0.3171,
       "step": 1275
     },
     {
-      "epoch": 0.1625,
-      "grad_norm": 5.599034786224365,
-      "learning_rate": 1.787466666666667e-05,
-      "loss": 0.3472,
+      "epoch": 0.13,
+      "grad_norm": 5.469560623168945,
+      "learning_rate": 2.290263157894737e-05,
+      "loss": 0.3374,
       "step": 1300
     },
     {
-      "epoch": 0.165625,
-      "grad_norm": 5.392984867095947,
-      "learning_rate": 1.7808e-05,
-      "loss": 0.3073,
+      "epoch": 0.1325,
+      "grad_norm": 5.263423919677734,
+      "learning_rate": 2.283684210526316e-05,
+      "loss": 0.2969,
       "step": 1325
     },
     {
-      "epoch": 0.16875,
-      "grad_norm": 4.690824508666992,
-      "learning_rate": 1.7741333333333336e-05,
-      "loss": 0.3096,
+      "epoch": 0.135,
+      "grad_norm": 4.6988983154296875,
+      "learning_rate": 2.2771052631578948e-05,
+      "loss": 0.3016,
       "step": 1350
     },
     {
-      "epoch": 0.171875,
-      "grad_norm": 5.487542152404785,
-      "learning_rate": 1.767466666666667e-05,
-      "loss": 0.2872,
+      "epoch": 0.1375,
+      "grad_norm": 5.226208209991455,
+      "learning_rate": 2.2705263157894737e-05,
+      "loss": 0.2781,
       "step": 1375
     },
     {
-      "epoch": 0.175,
-      "grad_norm": 5.0103440284729,
-      "learning_rate": 1.7608e-05,
-      "loss": 0.28,
+      "epoch": 0.14,
+      "grad_norm": 5.410859107971191,
+      "learning_rate": 2.2639473684210526e-05,
+      "loss": 0.2711,
       "step": 1400
     },
     {
-      "epoch": 0.178125,
-      "grad_norm": 4.374607563018799,
-      "learning_rate": 1.7541333333333334e-05,
-      "loss": 0.2907,
+      "epoch": 0.1425,
+      "grad_norm": 4.244476318359375,
+      "learning_rate": 2.2573684210526315e-05,
+      "loss": 0.2781,
       "step": 1425
     },
     {
-      "epoch": 0.18125,
-      "grad_norm": 3.8199336528778076,
-      "learning_rate": 1.7474666666666667e-05,
-      "loss": 0.2721,
+      "epoch": 0.145,
+      "grad_norm": 3.670611619949341,
+      "learning_rate": 2.2507894736842105e-05,
+      "loss": 0.2651,
       "step": 1450
     },
     {
-      "epoch": 0.184375,
-      "grad_norm": 3.263697862625122,
-      "learning_rate": 1.7408e-05,
-      "loss": 0.2018,
+      "epoch": 0.1475,
+      "grad_norm": 2.861996650695801,
+      "learning_rate": 2.2442105263157894e-05,
+      "loss": 0.1924,
       "step": 1475
     },
     {
-      "epoch": 0.1875,
-      "grad_norm": 3.686453104019165,
-      "learning_rate": 1.7341333333333333e-05,
-      "loss": 0.1804,
+      "epoch": 0.15,
+      "grad_norm": 3.5922746658325195,
+      "learning_rate": 2.2376315789473683e-05,
+      "loss": 0.1704,
       "step": 1500
     },
     {
-      "epoch": 0.190625,
-      "grad_norm": 3.122502088546753,
-      "learning_rate": 1.727466666666667e-05,
-      "loss": 0.1864,
+      "epoch": 0.1525,
+      "grad_norm": 2.853532552719116,
+      "learning_rate": 2.2310526315789472e-05,
+      "loss": 0.1766,
       "step": 1525
     },
     {
-      "epoch": 0.19375,
-      "grad_norm": 3.8203911781311035,
-      "learning_rate": 1.7208000000000002e-05,
-      "loss": 0.1822,
+      "epoch": 0.155,
+      "grad_norm": 4.346705913543701,
+      "learning_rate": 2.2244736842105264e-05,
+      "loss": 0.1725,
       "step": 1550
     },
     {
-      "epoch": 0.196875,
-      "grad_norm": 3.1615259647369385,
-      "learning_rate": 1.7141333333333335e-05,
-      "loss": 0.1607,
+      "epoch": 0.1575,
+      "grad_norm": 3.504809856414795,
+      "learning_rate": 2.2178947368421054e-05,
+      "loss": 0.1531,
       "step": 1575
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 3.136589288711548,
-      "learning_rate": 1.7074666666666668e-05,
-      "loss": 0.1681,
+      "epoch": 0.16,
+      "grad_norm": 3.0613338947296143,
+      "learning_rate": 2.2113157894736843e-05,
+      "loss": 0.1583,
       "step": 1600
     },
     {
-      "epoch": 0.203125,
-      "grad_norm": 5.022261142730713,
-      "learning_rate": 1.7008000000000004e-05,
-      "loss": 0.2476,
+      "epoch": 0.1625,
+      "grad_norm": 3.996514320373535,
+      "learning_rate": 2.2047368421052632e-05,
+      "loss": 0.2377,
       "step": 1625
     },
     {
-      "epoch": 0.20625,
-      "grad_norm": 4.3968377113342285,
-      "learning_rate": 1.6941333333333333e-05,
-      "loss": 0.2578,
+      "epoch": 0.165,
+      "grad_norm": 4.162107467651367,
+      "learning_rate": 2.198157894736842e-05,
+      "loss": 0.2492,
       "step": 1650
     },
     {
-      "epoch": 0.209375,
-      "grad_norm": 5.124329090118408,
-      "learning_rate": 1.687466666666667e-05,
-      "loss": 0.2646,
+      "epoch": 0.1675,
+      "grad_norm": 5.007922172546387,
+      "learning_rate": 2.1915789473684213e-05,
+      "loss": 0.2558,
       "step": 1675
     },
     {
-      "epoch": 0.2125,
-      "grad_norm": 3.7056658267974854,
-      "learning_rate": 1.6808000000000002e-05,
-      "loss": 0.251,
+      "epoch": 0.17,
+      "grad_norm": 3.683666944503784,
+      "learning_rate": 2.1850000000000003e-05,
+      "loss": 0.2413,
       "step": 1700
     },
     {
-      "epoch": 0.215625,
-      "grad_norm": 3.477151870727539,
-      "learning_rate": 1.6741333333333335e-05,
-      "loss": 0.1996,
+      "epoch": 0.1725,
+      "grad_norm": 3.344407320022583,
+      "learning_rate": 2.178421052631579e-05,
+      "loss": 0.1907,
       "step": 1725
     },
     {
-      "epoch": 0.21875,
-      "grad_norm": 3.3584837913513184,
-      "learning_rate": 1.6674666666666668e-05,
-      "loss": 0.1939,
+      "epoch": 0.175,
+      "grad_norm": 4.3516035079956055,
+      "learning_rate": 2.171842105263158e-05,
+      "loss": 0.1848,
       "step": 1750
     },
     {
-      "epoch": 0.221875,
-      "grad_norm": 3.136394739151001,
-      "learning_rate": 1.6608e-05,
-      "loss": 0.1664,
+      "epoch": 0.1775,
+      "grad_norm": 3.406522035598755,
+      "learning_rate": 2.165263157894737e-05,
+      "loss": 0.1558,
       "step": 1775
     },
     {
-      "epoch": 0.225,
-      "grad_norm": 2.997995376586914,
-      "learning_rate": 1.6541333333333334e-05,
-      "loss": 0.1731,
+      "epoch": 0.18,
+      "grad_norm": 3.0335240364074707,
+      "learning_rate": 2.158684210526316e-05,
+      "loss": 0.1641,
       "step": 1800
     },
     {
-      "epoch": 0.228125,
-      "grad_norm": 3.235027551651001,
-      "learning_rate": 1.6474666666666667e-05,
-      "loss": 0.1609,
+      "epoch": 0.1825,
+      "grad_norm": 2.9874353408813477,
+      "learning_rate": 2.1521052631578948e-05,
+      "loss": 0.1537,
       "step": 1825
     },
     {
-      "epoch": 0.23125,
-      "grad_norm": 2.657120704650879,
-      "learning_rate": 1.6408000000000003e-05,
-      "loss": 0.1563,
+      "epoch": 0.185,
+      "grad_norm": 2.5093448162078857,
+      "learning_rate": 2.1455263157894737e-05,
+      "loss": 0.149,
       "step": 1850
     },
     {
-      "epoch": 0.234375,
-      "grad_norm": 3.5700979232788086,
-      "learning_rate": 1.6341333333333336e-05,
-      "loss": 0.1554,
+      "epoch": 0.1875,
+      "grad_norm": 3.172337532043457,
+      "learning_rate": 2.138947368421053e-05,
+      "loss": 0.1456,
       "step": 1875
     },
     {
-      "epoch": 0.2375,
-      "grad_norm": 3.4939069747924805,
-      "learning_rate": 1.627466666666667e-05,
-      "loss": 0.1553,
+      "epoch": 0.19,
+      "grad_norm": 3.346684455871582,
+      "learning_rate": 2.132368421052632e-05,
+      "loss": 0.1464,
       "step": 1900
     },
     {
-      "epoch": 0.240625,
-      "grad_norm": 2.4021010398864746,
-      "learning_rate": 1.6208e-05,
-      "loss": 0.1494,
+      "epoch": 0.1925,
+      "grad_norm": 2.3370494842529297,
+      "learning_rate": 2.1257894736842108e-05,
+      "loss": 0.1414,
       "step": 1925
     },
     {
-      "epoch": 0.24375,
-      "grad_norm": 4.335009574890137,
-      "learning_rate": 1.6141333333333334e-05,
-      "loss": 0.2011,
+      "epoch": 0.195,
+      "grad_norm": 4.008378505706787,
+      "learning_rate": 2.1192105263157897e-05,
+      "loss": 0.1927,
       "step": 1950
     },
     {
-      "epoch": 0.246875,
-      "grad_norm": 4.650504112243652,
-      "learning_rate": 1.6074666666666667e-05,
-      "loss": 0.2702,
+      "epoch": 0.1975,
+      "grad_norm": 4.25955867767334,
+      "learning_rate": 2.1126315789473686e-05,
+      "loss": 0.2604,
       "step": 1975
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 4.410916328430176,
-      "learning_rate": 1.6008e-05,
-      "loss": 0.2607,
+      "epoch": 0.2,
+      "grad_norm": 4.2033305168151855,
+      "learning_rate": 2.1060526315789475e-05,
+      "loss": 0.2515,
       "step": 2000
     },
     {
-      "epoch": 0.25,
-      "eval_loss": 0.3708300292491913,
-      "eval_runtime": 74.5053,
-      "eval_samples_per_second": 28.24,
-      "eval_steps_per_second": 1.772,
-      "eval_wer": 19.865382817612414,
+      "epoch": 0.2,
+      "eval_loss": 0.4336438775062561,
+      "eval_runtime": 50.4135,
+      "eval_samples_per_second": 33.543,
+      "eval_steps_per_second": 2.103,
+      "eval_wer": 19.99500437117522,
       "step": 2000
     },
     {
-      "epoch": 0.253125,
-      "grad_norm": 3.3271749019622803,
-      "learning_rate": 1.5941333333333336e-05,
-      "loss": 0.2175,
+      "epoch": 0.2025,
+      "grad_norm": 3.966740131378174,
+      "learning_rate": 2.0994736842105265e-05,
+      "loss": 0.2106,
       "step": 2025
     },
     {
-      "epoch": 0.25625,
-      "grad_norm": 3.365081310272217,
-      "learning_rate": 1.5874666666666666e-05,
-      "loss": 0.1527,
+      "epoch": 0.205,
+      "grad_norm": 4.607847690582275,
+      "learning_rate": 2.0928947368421054e-05,
+      "loss": 0.1458,
       "step": 2050
     },
     {
-      "epoch": 0.259375,
-      "grad_norm": 3.701395273208618,
-      "learning_rate": 1.5808000000000002e-05,
-      "loss": 0.1461,
+      "epoch": 0.2075,
+      "grad_norm": 3.091517925262451,
+      "learning_rate": 2.0863157894736843e-05,
+      "loss": 0.1402,
       "step": 2075
     },
     {
-      "epoch": 0.2625,
-      "grad_norm": 2.8837661743164062,
-      "learning_rate": 1.5741333333333335e-05,
-      "loss": 0.1507,
+      "epoch": 0.21,
+      "grad_norm": 2.76373028755188,
+      "learning_rate": 2.0797368421052632e-05,
+      "loss": 0.1436,
       "step": 2100
     },
     {
-      "epoch": 0.265625,
-      "grad_norm": 3.2435319423675537,
-      "learning_rate": 1.5674666666666667e-05,
-      "loss": 0.1314,
+      "epoch": 0.2125,
+      "grad_norm": 3.131234884262085,
+      "learning_rate": 2.073157894736842e-05,
+      "loss": 0.1245,
       "step": 2125
     },
     {
-      "epoch": 0.26875,
-      "grad_norm": 2.9637367725372314,
-      "learning_rate": 1.5608e-05,
-      "loss": 0.1292,
+      "epoch": 0.215,
+      "grad_norm": 2.7513225078582764,
+      "learning_rate": 2.066578947368421e-05,
+      "loss": 0.1235,
       "step": 2150
     },
     {
-      "epoch": 0.271875,
-      "grad_norm": 3.535871982574463,
-      "learning_rate": 1.5541333333333337e-05,
-      "loss": 0.1332,
+      "epoch": 0.2175,
+      "grad_norm": 3.4547324180603027,
+      "learning_rate": 2.06e-05,
+      "loss": 0.1264,
       "step": 2175
     },
     {
-      "epoch": 0.275,
-      "grad_norm": 3.633970022201538,
-      "learning_rate": 1.547466666666667e-05,
-      "loss": 0.2285,
+      "epoch": 0.22,
+      "grad_norm": 3.542309522628784,
+      "learning_rate": 2.053421052631579e-05,
+      "loss": 0.2194,
       "step": 2200
     },
     {
-      "epoch": 0.278125,
-      "grad_norm": 4.483553409576416,
-      "learning_rate": 1.5408000000000002e-05,
-      "loss": 0.2325,
+      "epoch": 0.2225,
+      "grad_norm": 4.416931629180908,
+      "learning_rate": 2.046842105263158e-05,
+      "loss": 0.2266,
       "step": 2225
     },
     {
-      "epoch": 0.28125,
-      "grad_norm": 4.577600955963135,
-      "learning_rate": 1.5341333333333335e-05,
-      "loss": 0.256,
+      "epoch": 0.225,
+      "grad_norm": 4.454063415527344,
+      "learning_rate": 2.040263157894737e-05,
+      "loss": 0.2448,
       "step": 2250
     },
     {
-      "epoch": 0.284375,
-      "grad_norm": 3.32087779045105,
-      "learning_rate": 1.5274666666666668e-05,
-      "loss": 0.1758,
+      "epoch": 0.2275,
+      "grad_norm": 3.1339340209960938,
+      "learning_rate": 2.033684210526316e-05,
+      "loss": 0.1677,
       "step": 2275
     },
     {
-      "epoch": 0.2875,
-      "grad_norm": 3.1856462955474854,
-      "learning_rate": 1.5208e-05,
-      "loss": 0.1311,
+      "epoch": 0.23,
+      "grad_norm": 2.6697356700897217,
+      "learning_rate": 2.0271052631578948e-05,
+      "loss": 0.123,
       "step": 2300
     },
     {
-      "epoch": 0.290625,
-      "grad_norm": 3.373046636581421,
-      "learning_rate": 1.5141333333333335e-05,
-      "loss": 0.1197,
+      "epoch": 0.2325,
+      "grad_norm": 3.379664182662964,
+      "learning_rate": 2.0205263157894737e-05,
+      "loss": 0.1138,
       "step": 2325
     },
     {
-      "epoch": 0.29375,
-      "grad_norm": 3.019298553466797,
-      "learning_rate": 1.5074666666666668e-05,
-      "loss": 0.1243,
+      "epoch": 0.235,
+      "grad_norm": 2.677241325378418,
+      "learning_rate": 2.0139473684210527e-05,
+      "loss": 0.1169,
       "step": 2350
     },
     {
-      "epoch": 0.296875,
-      "grad_norm": 2.5749433040618896,
-      "learning_rate": 1.5008000000000001e-05,
-      "loss": 0.1302,
+      "epoch": 0.2375,
+      "grad_norm": 2.45304799079895,
+      "learning_rate": 2.0073684210526316e-05,
+      "loss": 0.1218,
       "step": 2375
     },
     {
-      "epoch": 0.3,
-      "grad_norm": 4.252561092376709,
-      "learning_rate": 1.4941333333333334e-05,
-      "loss": 0.1277,
+      "epoch": 0.24,
+      "grad_norm": 5.708258628845215,
+      "learning_rate": 2.0007894736842105e-05,
+      "loss": 0.1217,
       "step": 2400
     },
     {
-      "epoch": 0.303125,
-      "grad_norm": 2.421847343444824,
-      "learning_rate": 1.4874666666666668e-05,
-      "loss": 0.1352,
+      "epoch": 0.2425,
+      "grad_norm": 2.341142177581787,
+      "learning_rate": 1.9942105263157894e-05,
+      "loss": 0.1286,
       "step": 2425
     },
     {
-      "epoch": 0.30625,
-      "grad_norm": 2.247629165649414,
-      "learning_rate": 1.4808e-05,
-      "loss": 0.1286,
+      "epoch": 0.245,
+      "grad_norm": 2.293227434158325,
+      "learning_rate": 1.9876315789473683e-05,
+      "loss": 0.1213,
       "step": 2450
     },
     {
-      "epoch": 0.309375,
-      "grad_norm": 2.594142436981201,
-      "learning_rate": 1.4741333333333334e-05,
-      "loss": 0.1355,
+      "epoch": 0.2475,
+      "grad_norm": 2.5403645038604736,
+      "learning_rate": 1.9810526315789472e-05,
+      "loss": 0.1296,
       "step": 2475
     },
     {
-      "epoch": 0.3125,
-      "grad_norm": 3.3857474327087402,
-      "learning_rate": 1.4674666666666669e-05,
-      "loss": 0.1539,
+      "epoch": 0.25,
+      "grad_norm": 3.274874448776245,
+      "learning_rate": 1.9744736842105265e-05,
+      "loss": 0.1471,
       "step": 2500
     },
     {
-      "epoch": 0.315625,
-      "grad_norm": 3.906268835067749,
-      "learning_rate": 1.4608000000000001e-05,
-      "loss": 0.2039,
+      "epoch": 0.2525,
+      "grad_norm": 3.9522368907928467,
+      "learning_rate": 1.9678947368421054e-05,
+      "loss": 0.1988,
       "step": 2525
     },
     {
-      "epoch": 0.31875,
-      "grad_norm": 4.0277204513549805,
-      "learning_rate": 1.4541333333333334e-05,
-      "loss": 0.1978,
+      "epoch": 0.255,
+      "grad_norm": 3.700836658477783,
+      "learning_rate": 1.9613157894736843e-05,
+      "loss": 0.1886,
       "step": 2550
     },
     {
-      "epoch": 0.321875,
-      "grad_norm": 3.9858391284942627,
-      "learning_rate": 1.4474666666666669e-05,
-      "loss": 0.2316,
+      "epoch": 0.2575,
+      "grad_norm": 3.8419880867004395,
+      "learning_rate": 1.9547368421052632e-05,
+      "loss": 0.2244,
       "step": 2575
     },
     {
-      "epoch": 0.325,
-      "grad_norm": 2.84332537651062,
-      "learning_rate": 1.4408000000000002e-05,
-      "loss": 0.1719,
+      "epoch": 0.26,
+      "grad_norm": 2.85577130317688,
+      "learning_rate": 1.9481578947368425e-05,
+      "loss": 0.1648,
       "step": 2600
     },
     {
-      "epoch": 0.328125,
-      "grad_norm": 3.294312000274658,
-      "learning_rate": 1.4341333333333334e-05,
-      "loss": 0.1421,
+      "epoch": 0.2625,
+      "grad_norm": 3.4031693935394287,
+      "learning_rate": 1.9415789473684214e-05,
+      "loss": 0.1355,
       "step": 2625
     },
     {
-      "epoch": 0.33125,
-      "grad_norm": 2.738583564758301,
-      "learning_rate": 1.4274666666666667e-05,
-      "loss": 0.1297,
+      "epoch": 0.265,
+      "grad_norm": 2.5746467113494873,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 0.1225,
       "step": 2650
     },
     {
-      "epoch": 0.334375,
-      "grad_norm": 2.4573464393615723,
-      "learning_rate": 1.4208000000000002e-05,
-      "loss": 0.1223,
+      "epoch": 0.2675,
+      "grad_norm": 4.394095420837402,
+      "learning_rate": 1.9284210526315792e-05,
+      "loss": 0.1159,
       "step": 2675
     },
     {
-      "epoch": 0.3375,
-      "grad_norm": 2.678255319595337,
-      "learning_rate": 1.4141333333333333e-05,
-      "loss": 0.1249,
+      "epoch": 0.27,
+      "grad_norm": 2.3961875438690186,
+      "learning_rate": 1.921842105263158e-05,
+      "loss": 0.1191,
       "step": 2700
     },
     {
-      "epoch": 0.340625,
-      "grad_norm": 3.8852200508117676,
-      "learning_rate": 1.4074666666666668e-05,
-      "loss": 0.1456,
+      "epoch": 0.2725,
+      "grad_norm": 3.188430070877075,
+      "learning_rate": 1.915263157894737e-05,
+      "loss": 0.14,
       "step": 2725
     },
     {
-      "epoch": 0.34375,
-      "grad_norm": 3.630040407180786,
-      "learning_rate": 1.4008000000000002e-05,
-      "loss": 0.1466,
+      "epoch": 0.275,
+      "grad_norm": 3.4510505199432373,
+      "learning_rate": 1.908684210526316e-05,
+      "loss": 0.142,
       "step": 2750
     },
     {
-      "epoch": 0.346875,
-      "grad_norm": 4.463013648986816,
-      "learning_rate": 1.3941333333333333e-05,
-      "loss": 0.2719,
+      "epoch": 0.2775,
+      "grad_norm": 4.626753807067871,
+      "learning_rate": 1.902105263157895e-05,
+      "loss": 0.2631,
       "step": 2775
     },
     {
-      "epoch": 0.35,
-      "grad_norm": 4.058387279510498,
-      "learning_rate": 1.3874666666666668e-05,
-      "loss": 0.244,
+      "epoch": 0.28,
+      "grad_norm": 3.846059799194336,
+      "learning_rate": 1.8955263157894738e-05,
+      "loss": 0.2364,
       "step": 2800
     },
     {
-      "epoch": 0.353125,
-      "grad_norm": 4.477724552154541,
-      "learning_rate": 1.3808e-05,
-      "loss": 0.2144,
+      "epoch": 0.2825,
+      "grad_norm": 4.325892925262451,
+      "learning_rate": 1.8889473684210527e-05,
+      "loss": 0.2073,
       "step": 2825
     },
     {
-      "epoch": 0.35625,
-      "grad_norm": 3.583326578140259,
-      "learning_rate": 1.3741333333333335e-05,
-      "loss": 0.2157,
+      "epoch": 0.285,
+      "grad_norm": 3.4237799644470215,
+      "learning_rate": 1.8823684210526316e-05,
+      "loss": 0.2088,
       "step": 2850
     },
     {
-      "epoch": 0.359375,
-      "grad_norm": 3.8221640586853027,
-      "learning_rate": 1.3674666666666668e-05,
-      "loss": 0.2151,
+      "epoch": 0.2875,
+      "grad_norm": 3.6403746604919434,
+      "learning_rate": 1.8757894736842105e-05,
+      "loss": 0.2092,
       "step": 2875
     },
     {
-      "epoch": 0.3625,
-      "grad_norm": 3.6878933906555176,
-      "learning_rate": 1.3608e-05,
-      "loss": 0.2063,
+      "epoch": 0.29,
+      "grad_norm": 3.4176881313323975,
+      "learning_rate": 1.8692105263157897e-05,
+      "loss": 0.1984,
       "step": 2900
     },
     {
-      "epoch": 0.365625,
-      "grad_norm": 3.0994338989257812,
-      "learning_rate": 1.3541333333333335e-05,
-      "loss": 0.1613,
+      "epoch": 0.2925,
+      "grad_norm": 2.5348989963531494,
+      "learning_rate": 1.8626315789473687e-05,
+      "loss": 0.1549,
       "step": 2925
     },
     {
-      "epoch": 0.36875,
-      "grad_norm": 2.727104902267456,
-      "learning_rate": 1.3474666666666667e-05,
-      "loss": 0.1278,
+      "epoch": 0.295,
+      "grad_norm": 2.551673173904419,
+      "learning_rate": 1.8560526315789476e-05,
+      "loss": 0.1225,
       "step": 2950
     },
     {
-      "epoch": 0.371875,
-      "grad_norm": 3.0303940773010254,
-      "learning_rate": 1.3408000000000001e-05,
-      "loss": 0.1347,
+      "epoch": 0.2975,
+      "grad_norm": 2.939521074295044,
+      "learning_rate": 1.8494736842105265e-05,
+      "loss": 0.1277,
       "step": 2975
     },
     {
-      "epoch": 0.375,
-      "grad_norm": 3.6327948570251465,
-      "learning_rate": 1.3341333333333336e-05,
-      "loss": 0.1887,
+      "epoch": 0.3,
+      "grad_norm": 3.426952600479126,
+      "learning_rate": 1.8428947368421054e-05,
+      "loss": 0.1792,
       "step": 3000
     },
     {
-      "epoch": 0.375,
-      "eval_loss": 0.34539544582366943,
-      "eval_runtime": 75.5211,
-      "eval_samples_per_second": 27.86,
-      "eval_steps_per_second": 1.748,
-      "eval_wer": 18.39768159296999,
+      "epoch": 0.3,
+      "eval_loss": 0.405368447303772,
+      "eval_runtime": 50.9273,
+      "eval_samples_per_second": 33.204,
+      "eval_steps_per_second": 2.081,
+      "eval_wer": 17.640814287498436,
       "step": 3000
     },
     {
-      "epoch": 0.378125,
-      "grad_norm": 4.803223133087158,
-      "learning_rate": 1.3274666666666667e-05,
-      "loss": 0.2279,
+      "epoch": 0.3025,
+      "grad_norm": 4.513154983520508,
+      "learning_rate": 1.8363157894736843e-05,
+      "loss": 0.2199,
       "step": 3025
     },
     {
-      "epoch": 0.38125,
-      "grad_norm": 4.9647536277771,
-      "learning_rate": 1.3208000000000001e-05,
-      "loss": 0.2556,
+      "epoch": 0.305,
+      "grad_norm": 4.8893208503723145,
+      "learning_rate": 1.8297368421052632e-05,
+      "loss": 0.2479,
       "step": 3050
     },
     {
-      "epoch": 0.384375,
-      "grad_norm": 3.8229358196258545,
-      "learning_rate": 1.3141333333333334e-05,
-      "loss": 0.1795,
+      "epoch": 0.3075,
+      "grad_norm": 3.0891120433807373,
+      "learning_rate": 1.823157894736842e-05,
+      "loss": 0.1728,
       "step": 3075
     },
     {
-      "epoch": 0.3875,
-      "grad_norm": 3.2917191982269287,
-      "learning_rate": 1.3074666666666669e-05,
-      "loss": 0.1355,
+      "epoch": 0.31,
+      "grad_norm": 3.2200894355773926,
+      "learning_rate": 1.816578947368421e-05,
+      "loss": 0.1299,
       "step": 3100
     },
     {
-      "epoch": 0.390625,
-      "grad_norm": 2.797985076904297,
-      "learning_rate": 1.3008e-05,
-      "loss": 0.1243,
+      "epoch": 0.3125,
+      "grad_norm": 2.606245517730713,
+      "learning_rate": 1.81e-05,
+      "loss": 0.1181,
       "step": 3125
     },
     {
-      "epoch": 0.39375,
-      "grad_norm": 3.5542221069335938,
-      "learning_rate": 1.2941333333333334e-05,
-      "loss": 0.1494,
+      "epoch": 0.315,
+      "grad_norm": 3.5022218227386475,
+      "learning_rate": 1.803421052631579e-05,
+      "loss": 0.1426,
       "step": 3150
     },
     {
-      "epoch": 0.396875,
-      "grad_norm": 3.7407355308532715,
-      "learning_rate": 1.2874666666666669e-05,
-      "loss": 0.2271,
+      "epoch": 0.3175,
+      "grad_norm": 3.6949498653411865,
+      "learning_rate": 1.7968421052631578e-05,
+      "loss": 0.2194,
       "step": 3175
     },
     {
-      "epoch": 0.4,
-      "grad_norm": 4.250736713409424,
-      "learning_rate": 1.2808e-05,
-      "loss": 0.2051,
+      "epoch": 0.32,
+      "grad_norm": 4.26692533493042,
+      "learning_rate": 1.7902631578947367e-05,
+      "loss": 0.1984,
       "step": 3200
     },
     {
-      "epoch": 0.403125,
-      "grad_norm": 4.248854160308838,
-      "learning_rate": 1.2741333333333335e-05,
-      "loss": 0.2251,
+      "epoch": 0.3225,
+      "grad_norm": 4.208311080932617,
+      "learning_rate": 1.783684210526316e-05,
+      "loss": 0.2165,
       "step": 3225
     },
     {
-      "epoch": 0.40625,
-      "grad_norm": 3.6842331886291504,
-      "learning_rate": 1.2674666666666669e-05,
-      "loss": 0.1806,
+      "epoch": 0.325,
+      "grad_norm": 3.339872360229492,
+      "learning_rate": 1.777105263157895e-05,
+      "loss": 0.1719,
       "step": 3250
     },
     {
-      "epoch": 0.409375,
-      "grad_norm": 4.215803623199463,
-      "learning_rate": 1.2608e-05,
-      "loss": 0.1938,
+      "epoch": 0.3275,
+      "grad_norm": 4.028900623321533,
+      "learning_rate": 1.7705263157894738e-05,
+      "loss": 0.1824,
       "step": 3275
     },
     {
-      "epoch": 0.4125,
-      "grad_norm": 2.878873348236084,
-      "learning_rate": 1.2541333333333335e-05,
-      "loss": 0.211,
+      "epoch": 0.33,
+      "grad_norm": 2.662748098373413,
+      "learning_rate": 1.7639473684210527e-05,
+      "loss": 0.198,
       "step": 3300
     },
     {
-      "epoch": 0.415625,
-      "grad_norm": 3.9587783813476562,
-      "learning_rate": 1.2474666666666668e-05,
-      "loss": 0.1601,
+      "epoch": 0.3325,
+      "grad_norm": 3.946589231491089,
+      "learning_rate": 1.7573684210526316e-05,
+      "loss": 0.1525,
       "step": 3325
     },
     {
-      "epoch": 0.41875,
-      "grad_norm": 3.0218279361724854,
-      "learning_rate": 1.2408e-05,
-      "loss": 0.1236,
+      "epoch": 0.335,
+      "grad_norm": 2.9382238388061523,
+      "learning_rate": 1.7507894736842105e-05,
+      "loss": 0.1176,
       "step": 3350
     },
     {
-      "epoch": 0.421875,
-      "grad_norm": 2.6972806453704834,
-      "learning_rate": 1.2341333333333333e-05,
-      "loss": 0.1112,
+      "epoch": 0.3375,
+      "grad_norm": 2.6483423709869385,
+      "learning_rate": 1.7442105263157894e-05,
+      "loss": 0.1067,
       "step": 3375
     },
     {
-      "epoch": 0.425,
-      "grad_norm": 2.4185972213745117,
-      "learning_rate": 1.2274666666666668e-05,
-      "loss": 0.1212,
+      "epoch": 0.34,
+      "grad_norm": 2.244067907333374,
+      "learning_rate": 1.7376315789473687e-05,
+      "loss": 0.1166,
       "step": 3400
     },
     {
-      "epoch": 0.428125,
-      "grad_norm": 3.9205973148345947,
-      "learning_rate": 1.2208000000000002e-05,
-      "loss": 0.1225,
+      "epoch": 0.3425,
+      "grad_norm": 3.1684019565582275,
+      "learning_rate": 1.7310526315789476e-05,
+      "loss": 0.118,
       "step": 3425
     },
     {
-      "epoch": 0.43125,
-      "grad_norm": 2.392932176589966,
-      "learning_rate": 1.2141333333333334e-05,
-      "loss": 0.1167,
+      "epoch": 0.345,
+      "grad_norm": 1.9989653825759888,
+      "learning_rate": 1.7244736842105265e-05,
+      "loss": 0.1111,
       "step": 3450
     },
     {
-      "epoch": 0.434375,
-      "grad_norm": 4.024794101715088,
-      "learning_rate": 1.2074666666666668e-05,
-      "loss": 0.1057,
+      "epoch": 0.3475,
+      "grad_norm": 2.656865119934082,
+      "learning_rate": 1.7178947368421054e-05,
+      "loss": 0.1,
       "step": 3475
     },
     {
-      "epoch": 0.4375,
-      "grad_norm": 3.367401599884033,
-      "learning_rate": 1.2008000000000003e-05,
-      "loss": 0.1309,
+      "epoch": 0.35,
+      "grad_norm": 3.111891269683838,
+      "learning_rate": 1.7113157894736843e-05,
+      "loss": 0.1265,
       "step": 3500
     },
     {
-      "epoch": 0.440625,
-      "grad_norm": 2.7556755542755127,
-      "learning_rate": 1.1941333333333334e-05,
-      "loss": 0.1235,
+      "epoch": 0.3525,
+      "grad_norm": 2.762462615966797,
+      "learning_rate": 1.7047368421052632e-05,
+      "loss": 0.1153,
       "step": 3525
     },
     {
-      "epoch": 0.44375,
-      "grad_norm": 3.158759117126465,
-      "learning_rate": 1.1874666666666668e-05,
-      "loss": 0.1233,
+      "epoch": 0.355,
+      "grad_norm": 3.240023612976074,
+      "learning_rate": 1.698157894736842e-05,
+      "loss": 0.1182,
       "step": 3550
     },
     {
-      "epoch": 0.446875,
-      "grad_norm": 4.0478010177612305,
-      "learning_rate": 1.1808000000000001e-05,
-      "loss": 0.1892,
+      "epoch": 0.3575,
+      "grad_norm": 3.96018648147583,
+      "learning_rate": 1.6915789473684214e-05,
+      "loss": 0.1821,
       "step": 3575
     },
     {
-      "epoch": 0.45,
-      "grad_norm": 3.5508739948272705,
-      "learning_rate": 1.1741333333333334e-05,
-      "loss": 0.2186,
+      "epoch": 0.36,
+      "grad_norm": 3.344484567642212,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 0.2111,
       "step": 3600
     },
     {
-      "epoch": 0.453125,
-      "grad_norm": 3.6009671688079834,
-      "learning_rate": 1.1674666666666667e-05,
-      "loss": 0.2093,
+      "epoch": 0.3625,
+      "grad_norm": 3.38057804107666,
+      "learning_rate": 1.6784210526315792e-05,
+      "loss": 0.2019,
       "step": 3625
     },
     {
-      "epoch": 0.45625,
-      "grad_norm": 2.172722578048706,
-      "learning_rate": 1.1608000000000001e-05,
-      "loss": 0.1282,
+      "epoch": 0.365,
+      "grad_norm": 2.275235652923584,
+      "learning_rate": 1.671842105263158e-05,
+      "loss": 0.1221,
       "step": 3650
     },
     {
-      "epoch": 0.459375,
-      "grad_norm": 2.729567050933838,
-      "learning_rate": 1.1541333333333332e-05,
-      "loss": 0.1096,
+      "epoch": 0.3675,
+      "grad_norm": 4.385953903198242,
+      "learning_rate": 1.665263157894737e-05,
+      "loss": 0.1028,
       "step": 3675
     },
     {
-      "epoch": 0.4625,
-      "grad_norm": 2.7863428592681885,
-      "learning_rate": 1.1474666666666667e-05,
-      "loss": 0.1115,
+      "epoch": 0.37,
+      "grad_norm": 2.773892641067505,
+      "learning_rate": 1.658684210526316e-05,
+      "loss": 0.1061,
       "step": 3700
     },
     {
-      "epoch": 0.465625,
-      "grad_norm": 2.7164411544799805,
-      "learning_rate": 1.1408000000000002e-05,
-      "loss": 0.1365,
+      "epoch": 0.3725,
+      "grad_norm": 2.516683578491211,
+      "learning_rate": 1.652105263157895e-05,
+      "loss": 0.1309,
       "step": 3725
     },
     {
-      "epoch": 0.46875,
-      "grad_norm": 3.918790340423584,
-      "learning_rate": 1.1341333333333336e-05,
-      "loss": 0.1759,
+      "epoch": 0.375,
+      "grad_norm": 3.5930490493774414,
+      "learning_rate": 1.6455263157894738e-05,
+      "loss": 0.1644,
       "step": 3750
     },
     {
-      "epoch": 0.471875,
-      "grad_norm": 3.520095109939575,
-      "learning_rate": 1.1274666666666667e-05,
-      "loss": 0.2138,
+      "epoch": 0.3775,
+      "grad_norm": 3.509566068649292,
+      "learning_rate": 1.6389473684210527e-05,
+      "loss": 0.203,
       "step": 3775
     },
     {
-      "epoch": 0.475,
-      "grad_norm": 4.172083854675293,
-      "learning_rate": 1.1208000000000002e-05,
-      "loss": 0.2189,
+      "epoch": 0.38,
+      "grad_norm": 3.856156826019287,
+      "learning_rate": 1.6323684210526316e-05,
+      "loss": 0.2105,
       "step": 3800
     },
     {
-      "epoch": 0.478125,
-      "grad_norm": 4.076236724853516,
-      "learning_rate": 1.1141333333333335e-05,
-      "loss": 0.2079,
+      "epoch": 0.3825,
+      "grad_norm": 3.885843515396118,
+      "learning_rate": 1.6257894736842105e-05,
+      "loss": 0.1998,
       "step": 3825
     },
     {
-      "epoch": 0.48125,
-      "grad_norm": 4.950024604797363,
-      "learning_rate": 1.1074666666666667e-05,
-      "loss": 0.2233,
+      "epoch": 0.385,
+      "grad_norm": 4.575178146362305,
+      "learning_rate": 1.6192105263157894e-05,
+      "loss": 0.2117,
       "step": 3850
     },
     {
-      "epoch": 0.484375,
-      "grad_norm": 3.6588923931121826,
-      "learning_rate": 1.1008e-05,
-      "loss": 0.2011,
+      "epoch": 0.3875,
+      "grad_norm": 3.0911362171173096,
+      "learning_rate": 1.6126315789473683e-05,
+      "loss": 0.1898,
       "step": 3875
     },
     {
-      "epoch": 0.4875,
-      "grad_norm": 2.7259421348571777,
-      "learning_rate": 1.0941333333333335e-05,
-      "loss": 0.1419,
+      "epoch": 0.39,
+      "grad_norm": 2.5276999473571777,
+      "learning_rate": 1.6060526315789472e-05,
+      "loss": 0.1361,
       "step": 3900
     },
     {
-      "epoch": 0.490625,
-      "grad_norm": 4.260537147521973,
-      "learning_rate": 1.0874666666666666e-05,
-      "loss": 0.1221,
+      "epoch": 0.3925,
+      "grad_norm": 2.737013339996338,
+      "learning_rate": 1.5994736842105265e-05,
+      "loss": 0.1152,
       "step": 3925
     },
     {
-      "epoch": 0.49375,
-      "grad_norm": 2.9953160285949707,
-      "learning_rate": 1.0808e-05,
-      "loss": 0.1192,
+      "epoch": 0.395,
+      "grad_norm": 2.925593852996826,
+      "learning_rate": 1.5928947368421054e-05,
+      "loss": 0.1125,
       "step": 3950
     },
     {
-      "epoch": 0.496875,
-      "grad_norm": 5.537333011627197,
-      "learning_rate": 1.0741333333333335e-05,
-      "loss": 0.2057,
+      "epoch": 0.3975,
+      "grad_norm": 5.276342868804932,
+      "learning_rate": 1.5863157894736843e-05,
+      "loss": 0.1966,
       "step": 3975
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 4.265567302703857,
-      "learning_rate": 1.0674666666666666e-05,
-      "loss": 0.2607,
+      "epoch": 0.4,
+      "grad_norm": 3.9677958488464355,
+      "learning_rate": 1.5797368421052632e-05,
+      "loss": 0.2485,
       "step": 4000
     },
     {
-      "epoch": 0.5,
-      "eval_loss": 0.3218025863170624,
-      "eval_runtime": 75.7693,
-      "eval_samples_per_second": 27.768,
-      "eval_steps_per_second": 1.742,
-      "eval_wer": 16.808450967560997,
+      "epoch": 0.4,
+      "eval_loss": 0.3804095387458801,
+      "eval_runtime": 51.2349,
+      "eval_samples_per_second": 33.005,
+      "eval_steps_per_second": 2.069,
+      "eval_wer": 16.379418009241913,
       "step": 4000
     },
     {
-      "epoch": 0.503125,
-      "grad_norm": 2.978968620300293,
-      "learning_rate": 1.0608e-05,
-      "loss": 0.1951,
+      "epoch": 0.4025,
+      "grad_norm": 2.8115105628967285,
+      "learning_rate": 1.573157894736842e-05,
+      "loss": 0.1848,
       "step": 4025
     },
     {
-      "epoch": 0.50625,
-      "grad_norm": 5.042616367340088,
-      "learning_rate": 1.0541333333333335e-05,
-      "loss": 0.2567,
+      "epoch": 0.405,
+      "grad_norm": 4.65349006652832,
+      "learning_rate": 1.566578947368421e-05,
+      "loss": 0.2436,
       "step": 4050
     },
     {
-      "epoch": 0.509375,
-      "grad_norm": 4.18173885345459,
-      "learning_rate": 1.0474666666666668e-05,
-      "loss": 0.1932,
+      "epoch": 0.4075,
+      "grad_norm": 5.190550327301025,
+      "learning_rate": 1.56e-05,
+      "loss": 0.1812,
       "step": 4075
     },
     {
-      "epoch": 0.5125,
-      "grad_norm": 3.0428967475891113,
-      "learning_rate": 1.0408000000000001e-05,
-      "loss": 0.1743,
+      "epoch": 0.41,
+      "grad_norm": 2.84175181388855,
+      "learning_rate": 1.553421052631579e-05,
+      "loss": 0.1656,
       "step": 4100
     },
     {
-      "epoch": 0.515625,
-      "grad_norm": 2.8713204860687256,
-      "learning_rate": 1.0341333333333334e-05,
-      "loss": 0.1261,
+      "epoch": 0.4125,
+      "grad_norm": 2.3396739959716797,
+      "learning_rate": 1.5468421052631578e-05,
+      "loss": 0.1201,
       "step": 4125
     },
     {
-      "epoch": 0.51875,
-      "grad_norm": 2.912363052368164,
-      "learning_rate": 1.0274666666666668e-05,
-      "loss": 0.1112,
+      "epoch": 0.415,
+      "grad_norm": 2.581535816192627,
+      "learning_rate": 1.5402631578947367e-05,
+      "loss": 0.1069,
       "step": 4150
     },
     {
-      "epoch": 0.521875,
-      "grad_norm": 2.445664167404175,
-      "learning_rate": 1.0208e-05,
-      "loss": 0.1133,
+      "epoch": 0.4175,
+      "grad_norm": 2.381739616394043,
+      "learning_rate": 1.5336842105263156e-05,
+      "loss": 0.1083,
       "step": 4175
     },
     {
-      "epoch": 0.525,
-      "grad_norm": 2.2317187786102295,
-      "learning_rate": 1.0141333333333334e-05,
-      "loss": 0.106,
+      "epoch": 0.42,
+      "grad_norm": 2.072289228439331,
+      "learning_rate": 1.527105263157895e-05,
+      "loss": 0.1015,
       "step": 4200
     },
     {
-      "epoch": 0.528125,
-      "grad_norm": 2.4223077297210693,
-      "learning_rate": 1.0074666666666669e-05,
-      "loss": 0.1142,
+      "epoch": 0.4225,
+      "grad_norm": 2.175671339035034,
+      "learning_rate": 1.5205263157894736e-05,
+      "loss": 0.108,
       "step": 4225
     },
     {
-      "epoch": 0.53125,
-      "grad_norm": 2.8847713470458984,
-      "learning_rate": 1.0008e-05,
-      "loss": 0.1158,
+      "epoch": 0.425,
+      "grad_norm": 2.63728666305542,
+      "learning_rate": 1.5139473684210529e-05,
+      "loss": 0.1096,
       "step": 4250
     },
     {
-      "epoch": 0.534375,
-      "grad_norm": 3.4072630405426025,
-      "learning_rate": 9.941333333333334e-06,
-      "loss": 0.1283,
+      "epoch": 0.4275,
+      "grad_norm": 3.103864908218384,
+      "learning_rate": 1.5073684210526318e-05,
+      "loss": 0.1202,
       "step": 4275
     },
     {
-      "epoch": 0.5375,
-      "grad_norm": 4.455233573913574,
-      "learning_rate": 9.874666666666669e-06,
-      "loss": 0.1867,
+      "epoch": 0.43,
+      "grad_norm": 3.9897496700286865,
+      "learning_rate": 1.5007894736842107e-05,
+      "loss": 0.1781,
       "step": 4300
     },
     {
-      "epoch": 0.540625,
-      "grad_norm": 3.465684175491333,
-      "learning_rate": 9.808000000000002e-06,
-      "loss": 0.1681,
+      "epoch": 0.4325,
+      "grad_norm": 2.9400570392608643,
+      "learning_rate": 1.4942105263157896e-05,
+      "loss": 0.1601,
       "step": 4325
     },
     {
-      "epoch": 0.54375,
-      "grad_norm": 3.8950164318084717,
-      "learning_rate": 9.741333333333334e-06,
-      "loss": 0.1812,
+      "epoch": 0.435,
+      "grad_norm": 3.374852418899536,
+      "learning_rate": 1.4876315789473685e-05,
+      "loss": 0.1722,
       "step": 4350
     },
     {
-      "epoch": 0.546875,
-      "grad_norm": 2.216827630996704,
-      "learning_rate": 9.674666666666667e-06,
-      "loss": 0.1069,
+      "epoch": 0.4375,
+      "grad_norm": 2.0654988288879395,
+      "learning_rate": 1.4810526315789474e-05,
+      "loss": 0.1003,
       "step": 4375
     },
     {
-      "epoch": 0.55,
-      "grad_norm": 2.3940842151641846,
-      "learning_rate": 9.608e-06,
-      "loss": 0.095,
+      "epoch": 0.44,
+      "grad_norm": 2.2996861934661865,
+      "learning_rate": 1.4744736842105263e-05,
+      "loss": 0.0905,
       "step": 4400
     },
     {
-      "epoch": 0.553125,
-      "grad_norm": 2.4614291191101074,
-      "learning_rate": 9.541333333333335e-06,
-      "loss": 0.1019,
+      "epoch": 0.4425,
+      "grad_norm": 2.3257508277893066,
+      "learning_rate": 1.4678947368421053e-05,
+      "loss": 0.097,
       "step": 4425
     },
     {
-      "epoch": 0.55625,
-      "grad_norm": 2.891763925552368,
-      "learning_rate": 9.474666666666668e-06,
-      "loss": 0.12,
+      "epoch": 0.445,
+      "grad_norm": 2.8020572662353516,
+      "learning_rate": 1.4613157894736843e-05,
+      "loss": 0.1142,
       "step": 4450
     },
     {
-      "epoch": 0.559375,
-      "grad_norm": 2.791774272918701,
-      "learning_rate": 9.408e-06,
-      "loss": 0.1268,
+      "epoch": 0.4475,
+      "grad_norm": 2.4309985637664795,
+      "learning_rate": 1.4547368421052632e-05,
+      "loss": 0.1203,
       "step": 4475
     },
     {
-      "epoch": 0.5625,
-      "grad_norm": 2.9557909965515137,
-      "learning_rate": 9.341333333333335e-06,
-      "loss": 0.1164,
+      "epoch": 0.45,
+      "grad_norm": 2.849823474884033,
+      "learning_rate": 1.4481578947368422e-05,
+      "loss": 0.112,
       "step": 4500
     },
     {
-      "epoch": 0.565625,
-      "grad_norm": 3.381051540374756,
-      "learning_rate": 9.274666666666668e-06,
-      "loss": 0.1464,
+      "epoch": 0.4525,
+      "grad_norm": 3.266927480697632,
+      "learning_rate": 1.441578947368421e-05,
+      "loss": 0.1398,
       "step": 4525
     },
     {
-      "epoch": 0.56875,
-      "grad_norm": 3.789724588394165,
-      "learning_rate": 9.208e-06,
-      "loss": 0.1947,
+      "epoch": 0.455,
+      "grad_norm": 3.6707375049591064,
+      "learning_rate": 1.435e-05,
+      "loss": 0.187,
       "step": 4550
     },
     {
-      "epoch": 0.571875,
-      "grad_norm": 3.7860305309295654,
-      "learning_rate": 9.141333333333333e-06,
-      "loss": 0.1897,
+      "epoch": 0.4575,
+      "grad_norm": 3.624910593032837,
+      "learning_rate": 1.4284210526315789e-05,
+      "loss": 0.1804,
       "step": 4575
     },
     {
-      "epoch": 0.575,
-      "grad_norm": 4.125986576080322,
-      "learning_rate": 9.074666666666668e-06,
-      "loss": 0.2003,
+      "epoch": 0.46,
+      "grad_norm": 3.9688940048217773,
+      "learning_rate": 1.4218421052631581e-05,
+      "loss": 0.191,
       "step": 4600
     },
     {
-      "epoch": 0.578125,
-      "grad_norm": 4.029356002807617,
-      "learning_rate": 9.008e-06,
-      "loss": 0.2014,
+      "epoch": 0.4625,
+      "grad_norm": 3.886730909347534,
+      "learning_rate": 1.415263157894737e-05,
+      "loss": 0.1943,
       "step": 4625
     },
     {
-      "epoch": 0.58125,
-      "grad_norm": 4.662783622741699,
-      "learning_rate": 8.941333333333334e-06,
-      "loss": 0.1962,
+      "epoch": 0.465,
+      "grad_norm": 4.5468668937683105,
+      "learning_rate": 1.408684210526316e-05,
+      "loss": 0.1859,
       "step": 4650
     },
     {
-      "epoch": 0.584375,
-      "grad_norm": 3.978227138519287,
-      "learning_rate": 8.874666666666667e-06,
-      "loss": 0.1693,
+      "epoch": 0.4675,
+      "grad_norm": 3.7898447513580322,
+      "learning_rate": 1.4021052631578949e-05,
+      "loss": 0.1625,
       "step": 4675
     },
     {
-      "epoch": 0.5875,
-      "grad_norm": 2.98833966255188,
-      "learning_rate": 8.808000000000001e-06,
-      "loss": 0.1565,
+      "epoch": 0.47,
+      "grad_norm": 2.8314898014068604,
+      "learning_rate": 1.3955263157894738e-05,
+      "loss": 0.1476,
       "step": 4700
     },
     {
-      "epoch": 0.590625,
-      "grad_norm": 4.219015121459961,
-      "learning_rate": 8.741333333333334e-06,
-      "loss": 0.1687,
+      "epoch": 0.4725,
+      "grad_norm": 3.8690438270568848,
+      "learning_rate": 1.3889473684210527e-05,
+      "loss": 0.1574,
       "step": 4725
     },
     {
-      "epoch": 0.59375,
-      "grad_norm": 2.8378167152404785,
-      "learning_rate": 8.674666666666668e-06,
-      "loss": 0.1611,
+      "epoch": 0.475,
+      "grad_norm": 2.5614655017852783,
+      "learning_rate": 1.3823684210526316e-05,
+      "loss": 0.1524,
       "step": 4750
     },
     {
-      "epoch": 0.596875,
-      "grad_norm": 2.5076210498809814,
-      "learning_rate": 8.608000000000001e-06,
-      "loss": 0.1186,
+      "epoch": 0.4775,
+      "grad_norm": 2.484726667404175,
+      "learning_rate": 1.3757894736842105e-05,
+      "loss": 0.1123,
       "step": 4775
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 2.2282755374908447,
-      "learning_rate": 8.541333333333334e-06,
-      "loss": 0.1267,
+      "epoch": 0.48,
+      "grad_norm": 2.0394532680511475,
+      "learning_rate": 1.3692105263157894e-05,
+      "loss": 0.1204,
       "step": 4800
     },
     {
-      "epoch": 0.603125,
-      "grad_norm": 3.080812692642212,
-      "learning_rate": 8.474666666666667e-06,
-      "loss": 0.1112,
+      "epoch": 0.4825,
+      "grad_norm": 2.798093318939209,
+      "learning_rate": 1.3626315789473684e-05,
+      "loss": 0.1052,
       "step": 4825
     },
     {
-      "epoch": 0.60625,
-      "grad_norm": 3.513218641281128,
-      "learning_rate": 8.408e-06,
-      "loss": 0.1326,
+      "epoch": 0.485,
+      "grad_norm": 2.972621202468872,
+      "learning_rate": 1.3560526315789473e-05,
+      "loss": 0.1216,
       "step": 4850
     },
     {
-      "epoch": 0.609375,
-      "grad_norm": 3.9359219074249268,
-      "learning_rate": 8.341333333333334e-06,
-      "loss": 0.1454,
+      "epoch": 0.4875,
+      "grad_norm": 3.6155519485473633,
+      "learning_rate": 1.3494736842105264e-05,
+      "loss": 0.1319,
       "step": 4875
     },
     {
-      "epoch": 0.6125,
-      "grad_norm": 3.585268259048462,
-      "learning_rate": 8.274666666666667e-06,
-      "loss": 0.1583,
+      "epoch": 0.49,
+      "grad_norm": 3.560375452041626,
+      "learning_rate": 1.3428947368421053e-05,
+      "loss": 0.1482,
       "step": 4900
     },
     {
-      "epoch": 0.615625,
-      "grad_norm": 3.322193145751953,
-      "learning_rate": 8.208e-06,
-      "loss": 0.1417,
+      "epoch": 0.4925,
+      "grad_norm": 2.468935489654541,
+      "learning_rate": 1.3363157894736843e-05,
+      "loss": 0.1345,
       "step": 4925
     },
     {
-      "epoch": 0.61875,
-      "grad_norm": 2.7378623485565186,
-      "learning_rate": 8.141333333333335e-06,
-      "loss": 0.1164,
+      "epoch": 0.495,
+      "grad_norm": 2.6389522552490234,
+      "learning_rate": 1.3297368421052633e-05,
+      "loss": 0.1115,
       "step": 4950
     },
     {
-      "epoch": 0.621875,
-      "grad_norm": 5.096762657165527,
-      "learning_rate": 8.074666666666667e-06,
-      "loss": 0.1077,
+      "epoch": 0.4975,
+      "grad_norm": 4.905831813812256,
+      "learning_rate": 1.3231578947368423e-05,
+      "loss": 0.103,
       "step": 4975
     },
     {
-      "epoch": 0.625,
-      "grad_norm": 3.030876636505127,
-      "learning_rate": 8.008e-06,
-      "loss": 0.106,
+      "epoch": 0.5,
+      "grad_norm": 2.904928207397461,
+      "learning_rate": 1.3165789473684213e-05,
+      "loss": 0.1007,
       "step": 5000
     },
     {
-      "epoch": 0.625,
-      "eval_loss": 0.32894453406333923,
-      "eval_runtime": 75.4732,
-      "eval_samples_per_second": 27.877,
-      "eval_steps_per_second": 1.749,
-      "eval_wer": 15.780125268766945,
+      "epoch": 0.5,
+      "eval_loss": 0.4055747985839844,
+      "eval_runtime": 51.151,
+      "eval_samples_per_second": 33.059,
+      "eval_steps_per_second": 2.072,
+      "eval_wer": 15.255401523666793,
       "step": 5000
     },
     {
-      "epoch": 0.628125,
-      "grad_norm": 3.1544792652130127,
-      "learning_rate": 7.941333333333335e-06,
-      "loss": 0.1599,
+      "epoch": 0.5025,
+      "grad_norm": 2.8446755409240723,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.1513,
       "step": 5025
     },
     {
-      "epoch": 0.63125,
-      "grad_norm": 3.2215702533721924,
-      "learning_rate": 7.874666666666668e-06,
-      "loss": 0.1461,
+      "epoch": 0.505,
+      "grad_norm": 3.151038408279419,
+      "learning_rate": 1.303421052631579e-05,
+      "loss": 0.1359,
       "step": 5050
     },
     {
-      "epoch": 0.634375,
-      "grad_norm": 3.7183897495269775,
-      "learning_rate": 7.808e-06,
-      "loss": 0.2221,
+      "epoch": 0.5075,
+      "grad_norm": 3.524214744567871,
+      "learning_rate": 1.296842105263158e-05,
+      "loss": 0.2112,
       "step": 5075
     },
     {
-      "epoch": 0.6375,
-      "grad_norm": 3.5974223613739014,
-      "learning_rate": 7.741333333333333e-06,
-      "loss": 0.2,
+      "epoch": 0.51,
+      "grad_norm": 3.5379486083984375,
+      "learning_rate": 1.2902631578947369e-05,
+      "loss": 0.1911,
       "step": 5100
     },
     {
-      "epoch": 0.640625,
-      "grad_norm": 3.8318378925323486,
-      "learning_rate": 7.674666666666666e-06,
-      "loss": 0.1743,
+      "epoch": 0.5125,
+      "grad_norm": 3.4644596576690674,
+      "learning_rate": 1.2836842105263158e-05,
+      "loss": 0.1659,
       "step": 5125
     },
     {
-      "epoch": 0.64375,
-      "grad_norm": 4.3595290184021,
-      "learning_rate": 7.608000000000001e-06,
-      "loss": 0.174,
+      "epoch": 0.515,
+      "grad_norm": 4.6146464347839355,
+      "learning_rate": 1.2771052631578947e-05,
+      "loss": 0.1658,
       "step": 5150
     },
     {
-      "epoch": 0.646875,
-      "grad_norm": 2.820388078689575,
-      "learning_rate": 7.5413333333333335e-06,
-      "loss": 0.1415,
+      "epoch": 0.5175,
+      "grad_norm": 2.5683515071868896,
+      "learning_rate": 1.2705263157894736e-05,
+      "loss": 0.1355,
       "step": 5175
     },
     {
-      "epoch": 0.65,
-      "grad_norm": 2.1610324382781982,
-      "learning_rate": 7.474666666666666e-06,
-      "loss": 0.0955,
+      "epoch": 0.52,
+      "grad_norm": 2.089280843734741,
+      "learning_rate": 1.2639473684210525e-05,
+      "loss": 0.0911,
       "step": 5200
     },
     {
-      "epoch": 0.653125,
-      "grad_norm": 2.079317092895508,
-      "learning_rate": 7.408000000000001e-06,
-      "loss": 0.1016,
+      "epoch": 0.5225,
+      "grad_norm": 1.9280669689178467,
+      "learning_rate": 1.2573684210526315e-05,
+      "loss": 0.0963,
       "step": 5225
     },
     {
-      "epoch": 0.65625,
-      "grad_norm": 2.3071603775024414,
-      "learning_rate": 7.341333333333334e-06,
-      "loss": 0.1042,
+      "epoch": 0.525,
+      "grad_norm": 2.2376904487609863,
+      "learning_rate": 1.2507894736842104e-05,
+      "loss": 0.0989,
       "step": 5250
     },
     {
-      "epoch": 0.659375,
-      "grad_norm": 2.3762526512145996,
-      "learning_rate": 7.2746666666666674e-06,
-      "loss": 0.1058,
+      "epoch": 0.5275,
+      "grad_norm": 2.2531654834747314,
+      "learning_rate": 1.2442105263157895e-05,
+      "loss": 0.1003,
       "step": 5275
     },
     {
-      "epoch": 0.6625,
-      "grad_norm": 3.6836395263671875,
-      "learning_rate": 7.208e-06,
-      "loss": 0.1087,
+      "epoch": 0.53,
+      "grad_norm": 3.8537986278533936,
+      "learning_rate": 1.2376315789473684e-05,
+      "loss": 0.1044,
       "step": 5300
     },
     {
-      "epoch": 0.665625,
-      "grad_norm": 3.0931732654571533,
-      "learning_rate": 7.141333333333333e-06,
-      "loss": 0.1062,
+      "epoch": 0.5325,
+      "grad_norm": 2.8962442874908447,
+      "learning_rate": 1.2310526315789474e-05,
+      "loss": 0.1009,
       "step": 5325
     },
     {
-      "epoch": 0.66875,
-      "grad_norm": 4.019095420837402,
-      "learning_rate": 7.074666666666668e-06,
-      "loss": 0.1453,
+      "epoch": 0.535,
+      "grad_norm": 4.504721641540527,
+      "learning_rate": 1.2244736842105264e-05,
+      "loss": 0.1372,
       "step": 5350
     },
     {
-      "epoch": 0.671875,
-      "grad_norm": 3.419175386428833,
-      "learning_rate": 7.0080000000000005e-06,
-      "loss": 0.1721,
+      "epoch": 0.5375,
+      "grad_norm": 3.3476779460906982,
+      "learning_rate": 1.2178947368421054e-05,
+      "loss": 0.166,
       "step": 5375
     },
     {
-      "epoch": 0.675,
-      "grad_norm": 3.387830972671509,
-      "learning_rate": 6.941333333333334e-06,
-      "loss": 0.1652,
+      "epoch": 0.54,
+      "grad_norm": 3.1945645809173584,
+      "learning_rate": 1.2113157894736844e-05,
+      "loss": 0.1565,
       "step": 5400
     },
     {
-      "epoch": 0.678125,
-      "grad_norm": 3.58986234664917,
-      "learning_rate": 6.874666666666667e-06,
-      "loss": 0.1639,
+      "epoch": 0.5425,
+      "grad_norm": 3.339404344558716,
+      "learning_rate": 1.2047368421052633e-05,
+      "loss": 0.1515,
       "step": 5425
     },
     {
-      "epoch": 0.68125,
-      "grad_norm": 4.178884506225586,
-      "learning_rate": 6.808e-06,
-      "loss": 0.1533,
+      "epoch": 0.545,
+      "grad_norm": 3.698160409927368,
+      "learning_rate": 1.1981578947368422e-05,
+      "loss": 0.1367,
       "step": 5450
     },
     {
-      "epoch": 0.684375,
-      "grad_norm": 3.2337114810943604,
-      "learning_rate": 6.741333333333334e-06,
-      "loss": 0.1237,
+      "epoch": 0.5475,
+      "grad_norm": 2.921149730682373,
+      "learning_rate": 1.1915789473684211e-05,
+      "loss": 0.1084,
       "step": 5475
     },
     {
-      "epoch": 0.6875,
-      "grad_norm": 2.892301321029663,
-      "learning_rate": 6.674666666666667e-06,
-      "loss": 0.1215,
+      "epoch": 0.55,
+      "grad_norm": 2.548079013824463,
+      "learning_rate": 1.185e-05,
+      "loss": 0.1104,
       "step": 5500
     },
     {
-      "epoch": 0.690625,
-      "grad_norm": 4.553407669067383,
-      "learning_rate": 6.608000000000001e-06,
-      "loss": 0.1537,
+      "epoch": 0.5525,
+      "grad_norm": 4.426082134246826,
+      "learning_rate": 1.1784210526315791e-05,
+      "loss": 0.1455,
       "step": 5525
     },
     {
-      "epoch": 0.69375,
-      "grad_norm": 3.8401100635528564,
-      "learning_rate": 6.541333333333334e-06,
-      "loss": 0.1816,
+      "epoch": 0.555,
+      "grad_norm": 3.9994497299194336,
+      "learning_rate": 1.171842105263158e-05,
+      "loss": 0.1679,
       "step": 5550
     },
     {
-      "epoch": 0.696875,
-      "grad_norm": 2.8084216117858887,
-      "learning_rate": 6.474666666666667e-06,
-      "loss": 0.1381,
+      "epoch": 0.5575,
+      "grad_norm": 2.904689073562622,
+      "learning_rate": 1.1652631578947369e-05,
+      "loss": 0.1271,
       "step": 5575
     },
     {
-      "epoch": 0.7,
-      "grad_norm": 2.182170867919922,
-      "learning_rate": 6.408000000000001e-06,
-      "loss": 0.0979,
+      "epoch": 0.56,
+      "grad_norm": 2.0764057636260986,
+      "learning_rate": 1.1586842105263158e-05,
+      "loss": 0.0852,
       "step": 5600
     },
     {
-      "epoch": 0.703125,
-      "grad_norm": 3.2050559520721436,
-      "learning_rate": 6.341333333333334e-06,
-      "loss": 0.0822,
+      "epoch": 0.5625,
+      "grad_norm": 2.4948599338531494,
+      "learning_rate": 1.1521052631578947e-05,
+      "loss": 0.0706,
       "step": 5625
     },
     {
-      "epoch": 0.70625,
-      "grad_norm": 2.4150376319885254,
-      "learning_rate": 6.274666666666667e-06,
-      "loss": 0.0815,
+      "epoch": 0.565,
+      "grad_norm": 2.049166440963745,
+      "learning_rate": 1.1455263157894736e-05,
+      "loss": 0.0725,
       "step": 5650
     },
     {
-      "epoch": 0.709375,
-      "grad_norm": 2.0708541870117188,
-      "learning_rate": 6.2080000000000005e-06,
-      "loss": 0.0944,
+      "epoch": 0.5675,
+      "grad_norm": 1.8661171197891235,
+      "learning_rate": 1.1389473684210526e-05,
+      "loss": 0.0896,
       "step": 5675
     },
     {
-      "epoch": 0.7125,
-      "grad_norm": 2.932088851928711,
-      "learning_rate": 6.141333333333333e-06,
-      "loss": 0.1024,
+      "epoch": 0.57,
+      "grad_norm": 2.9094321727752686,
+      "learning_rate": 1.1323684210526316e-05,
+      "loss": 0.0965,
       "step": 5700
     },
     {
-      "epoch": 0.715625,
-      "grad_norm": 2.245450258255005,
-      "learning_rate": 6.074666666666668e-06,
-      "loss": 0.101,
+      "epoch": 0.5725,
+      "grad_norm": 2.1191153526306152,
+      "learning_rate": 1.1257894736842106e-05,
+      "loss": 0.0959,
       "step": 5725
     },
     {
-      "epoch": 0.71875,
-      "grad_norm": 2.2716262340545654,
-      "learning_rate": 6.008000000000001e-06,
-      "loss": 0.1,
+      "epoch": 0.575,
+      "grad_norm": 2.010484218597412,
+      "learning_rate": 1.1192105263157895e-05,
+      "loss": 0.0948,
       "step": 5750
     },
     {
-      "epoch": 0.721875,
-      "grad_norm": 2.496361494064331,
-      "learning_rate": 5.941333333333334e-06,
-      "loss": 0.0966,
+      "epoch": 0.5775,
+      "grad_norm": 2.4947898387908936,
+      "learning_rate": 1.1126315789473685e-05,
+      "loss": 0.0912,
       "step": 5775
     },
     {
-      "epoch": 0.725,
-      "grad_norm": 3.3539814949035645,
-      "learning_rate": 5.874666666666667e-06,
-      "loss": 0.0911,
+      "epoch": 0.58,
+      "grad_norm": 2.372864246368408,
+      "learning_rate": 1.1060526315789475e-05,
+      "loss": 0.0858,
       "step": 5800
     },
     {
-      "epoch": 0.728125,
-      "grad_norm": 2.1496963500976562,
-      "learning_rate": 5.808e-06,
-      "loss": 0.0799,
+      "epoch": 0.5825,
+      "grad_norm": 2.256497383117676,
+      "learning_rate": 1.0994736842105264e-05,
+      "loss": 0.0769,
       "step": 5825
     },
     {
-      "epoch": 0.73125,
-      "grad_norm": 2.5061728954315186,
-      "learning_rate": 5.741333333333335e-06,
-      "loss": 0.0877,
+      "epoch": 0.585,
+      "grad_norm": 2.3207082748413086,
+      "learning_rate": 1.0928947368421053e-05,
+      "loss": 0.0819,
       "step": 5850
     },
     {
-      "epoch": 0.734375,
-      "grad_norm": 2.4256293773651123,
-      "learning_rate": 5.6746666666666675e-06,
-      "loss": 0.0945,
+      "epoch": 0.5875,
+      "grad_norm": 2.424302339553833,
+      "learning_rate": 1.0863157894736844e-05,
+      "loss": 0.0881,
       "step": 5875
     },
     {
-      "epoch": 0.7375,
-      "grad_norm": 3.3995237350463867,
-      "learning_rate": 5.608e-06,
-      "loss": 0.1245,
+      "epoch": 0.59,
+      "grad_norm": 3.049173355102539,
+      "learning_rate": 1.0797368421052633e-05,
+      "loss": 0.1156,
       "step": 5900
     },
     {
-      "epoch": 0.740625,
-      "grad_norm": 4.556021213531494,
-      "learning_rate": 5.541333333333334e-06,
-      "loss": 0.181,
+      "epoch": 0.5925,
+      "grad_norm": 4.130237579345703,
+      "learning_rate": 1.0731578947368422e-05,
+      "loss": 0.1675,
       "step": 5925
     },
     {
-      "epoch": 0.74375,
-      "grad_norm": 3.8894693851470947,
-      "learning_rate": 5.474666666666667e-06,
-      "loss": 0.1692,
+      "epoch": 0.595,
+      "grad_norm": 3.366295337677002,
+      "learning_rate": 1.0665789473684211e-05,
+      "loss": 0.1566,
       "step": 5950
     },
     {
-      "epoch": 0.746875,
-      "grad_norm": 3.464264392852783,
-      "learning_rate": 5.408e-06,
-      "loss": 0.1457,
+      "epoch": 0.5975,
+      "grad_norm": 3.341735601425171,
+      "learning_rate": 1.06e-05,
+      "loss": 0.1339,
       "step": 5975
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 3.351585865020752,
-      "learning_rate": 5.341333333333334e-06,
-      "loss": 0.1376,
+      "epoch": 0.6,
+      "grad_norm": 3.5875508785247803,
+      "learning_rate": 1.053421052631579e-05,
+      "loss": 0.1296,
       "step": 6000
     },
     {
-      "epoch": 0.75,
-      "eval_loss": 0.3052073121070862,
-      "eval_runtime": 73.9486,
-      "eval_samples_per_second": 28.452,
-      "eval_steps_per_second": 1.785,
-      "eval_wer": 15.02757782555857,
+      "epoch": 0.6,
+      "eval_loss": 0.373067170381546,
+      "eval_runtime": 50.8444,
+      "eval_samples_per_second": 33.258,
+      "eval_steps_per_second": 2.085,
+      "eval_wer": 15.324091420007493,
       "step": 6000
     },
     {
-      "epoch": 0.753125,
-      "grad_norm": 3.405545473098755,
-      "learning_rate": 5.274666666666667e-06,
-      "loss": 0.1327,
+      "epoch": 0.6025,
+      "grad_norm": 3.2236549854278564,
+      "learning_rate": 1.046842105263158e-05,
+      "loss": 0.1232,
       "step": 6025
     },
     {
-      "epoch": 0.75625,
-      "grad_norm": 2.5915920734405518,
-      "learning_rate": 5.208000000000001e-06,
-      "loss": 0.1145,
+      "epoch": 0.605,
+      "grad_norm": 2.3574883937835693,
+      "learning_rate": 1.040263157894737e-05,
+      "loss": 0.1062,
       "step": 6050
     },
     {
-      "epoch": 0.759375,
-      "grad_norm": 2.2758867740631104,
-      "learning_rate": 5.141333333333334e-06,
-      "loss": 0.1053,
+      "epoch": 0.6075,
+      "grad_norm": 2.152470350265503,
+      "learning_rate": 1.0336842105263158e-05,
+      "loss": 0.0983,
       "step": 6075
     },
     {
-      "epoch": 0.7625,
-      "grad_norm": 2.8261423110961914,
-      "learning_rate": 5.0746666666666665e-06,
-      "loss": 0.0942,
+      "epoch": 0.61,
+      "grad_norm": 2.31156849861145,
+      "learning_rate": 1.0271052631578947e-05,
+      "loss": 0.0896,
       "step": 6100
     },
     {
-      "epoch": 0.765625,
-      "grad_norm": 3.362257480621338,
-      "learning_rate": 5.008000000000001e-06,
-      "loss": 0.0969,
+      "epoch": 0.6125,
+      "grad_norm": 3.150630474090576,
+      "learning_rate": 1.0205263157894737e-05,
+      "loss": 0.0922,
       "step": 6125
     },
     {
-      "epoch": 0.76875,
-      "grad_norm": 3.890949249267578,
-      "learning_rate": 4.941333333333334e-06,
-      "loss": 0.1068,
+      "epoch": 0.615,
+      "grad_norm": 3.061333656311035,
+      "learning_rate": 1.0139473684210526e-05,
+      "loss": 0.1011,
       "step": 6150
     },
     {
-      "epoch": 0.771875,
-      "grad_norm": 3.03787899017334,
-      "learning_rate": 4.874666666666667e-06,
-      "loss": 0.1094,
+      "epoch": 0.6175,
+      "grad_norm": 2.7992866039276123,
+      "learning_rate": 1.0073684210526317e-05,
+      "loss": 0.1027,
       "step": 6175
     },
     {
-      "epoch": 0.775,
-      "grad_norm": 2.8833038806915283,
-      "learning_rate": 4.808e-06,
-      "loss": 0.1043,
+      "epoch": 0.62,
+      "grad_norm": 2.506035089492798,
+      "learning_rate": 1.0007894736842106e-05,
+      "loss": 0.0981,
       "step": 6200
     },
     {
-      "epoch": 0.778125,
-      "grad_norm": 3.1083550453186035,
-      "learning_rate": 4.741333333333334e-06,
-      "loss": 0.1061,
+      "epoch": 0.6225,
+      "grad_norm": 2.611304998397827,
+      "learning_rate": 9.942105263157896e-06,
+      "loss": 0.0994,
       "step": 6225
     },
     {
-      "epoch": 0.78125,
-      "grad_norm": 3.4954771995544434,
-      "learning_rate": 4.674666666666667e-06,
-      "loss": 0.1013,
+      "epoch": 0.625,
+      "grad_norm": 3.2212979793548584,
+      "learning_rate": 9.876315789473686e-06,
+      "loss": 0.0942,
       "step": 6250
     },
     {
-      "epoch": 0.784375,
-      "grad_norm": 3.035095691680908,
-      "learning_rate": 4.608000000000001e-06,
-      "loss": 0.1119,
+      "epoch": 0.6275,
+      "grad_norm": 2.89421010017395,
+      "learning_rate": 9.810526315789475e-06,
+      "loss": 0.1061,
       "step": 6275
     },
     {
-      "epoch": 0.7875,
-      "grad_norm": 3.62898850440979,
-      "learning_rate": 4.5413333333333334e-06,
-      "loss": 0.1253,
+      "epoch": 0.63,
+      "grad_norm": 3.2011821269989014,
+      "learning_rate": 9.744736842105264e-06,
+      "loss": 0.1173,
       "step": 6300
     },
     {
-      "epoch": 0.790625,
-      "grad_norm": 2.595010280609131,
-      "learning_rate": 4.474666666666667e-06,
-      "loss": 0.1681,
+      "epoch": 0.6325,
+      "grad_norm": 2.451089859008789,
+      "learning_rate": 9.678947368421053e-06,
+      "loss": 0.1572,
       "step": 6325
     },
     {
-      "epoch": 0.79375,
-      "grad_norm": 3.7245900630950928,
-      "learning_rate": 4.408000000000001e-06,
-      "loss": 0.1501,
+      "epoch": 0.635,
+      "grad_norm": 3.488110303878784,
+      "learning_rate": 9.613157894736842e-06,
+      "loss": 0.1406,
       "step": 6350
     },
     {
-      "epoch": 0.796875,
-      "grad_norm": 2.7315571308135986,
-      "learning_rate": 4.344e-06,
-      "loss": 0.1395,
+      "epoch": 0.6375,
+      "grad_norm": 2.5751559734344482,
+      "learning_rate": 9.55e-06,
+      "loss": 0.1297,
       "step": 6375
     },
     {
-      "epoch": 0.8,
-      "grad_norm": 2.3649332523345947,
-      "learning_rate": 4.277333333333334e-06,
-      "loss": 0.1111,
+      "epoch": 0.64,
+      "grad_norm": 2.2660510540008545,
+      "learning_rate": 9.48421052631579e-06,
+      "loss": 0.1049,
       "step": 6400
     },
     {
-      "epoch": 0.803125,
-      "grad_norm": 2.491359233856201,
-      "learning_rate": 4.210666666666667e-06,
-      "loss": 0.1024,
+      "epoch": 0.6425,
+      "grad_norm": 2.2303407192230225,
+      "learning_rate": 9.41842105263158e-06,
+      "loss": 0.0953,
       "step": 6425
     },
     {
-      "epoch": 0.80625,
-      "grad_norm": 2.5985302925109863,
-      "learning_rate": 4.1440000000000005e-06,
-      "loss": 0.0925,
+      "epoch": 0.645,
+      "grad_norm": 2.3398027420043945,
+      "learning_rate": 9.352631578947369e-06,
+      "loss": 0.0856,
       "step": 6450
     },
     {
-      "epoch": 0.809375,
-      "grad_norm": 4.008167266845703,
-      "learning_rate": 4.077333333333333e-06,
-      "loss": 0.1385,
+      "epoch": 0.6475,
+      "grad_norm": 3.764880657196045,
+      "learning_rate": 9.286842105263158e-06,
+      "loss": 0.1303,
       "step": 6475
     },
     {
-      "epoch": 0.8125,
-      "grad_norm": 2.743041753768921,
-      "learning_rate": 4.010666666666667e-06,
-      "loss": 0.1289,
+      "epoch": 0.65,
+      "grad_norm": 2.613056182861328,
+      "learning_rate": 9.221052631578949e-06,
+      "loss": 0.1217,
       "step": 6500
     },
     {
-      "epoch": 0.815625,
-      "grad_norm": 4.4984893798828125,
-      "learning_rate": 3.944e-06,
-      "loss": 0.1709,
+      "epoch": 0.6525,
+      "grad_norm": 4.344733715057373,
+      "learning_rate": 9.155263157894738e-06,
+      "loss": 0.1597,
       "step": 6525
     },
     {
-      "epoch": 0.81875,
-      "grad_norm": 3.432147741317749,
-      "learning_rate": 3.8773333333333335e-06,
-      "loss": 0.1563,
+      "epoch": 0.655,
+      "grad_norm": 3.09989070892334,
+      "learning_rate": 9.089473684210527e-06,
+      "loss": 0.1456,
       "step": 6550
     },
     {
-      "epoch": 0.821875,
-      "grad_norm": 3.6097943782806396,
-      "learning_rate": 3.810666666666667e-06,
-      "loss": 0.159,
+      "epoch": 0.6575,
+      "grad_norm": 3.1914069652557373,
+      "learning_rate": 9.023684210526316e-06,
+      "loss": 0.1458,
       "step": 6575
     },
     {
-      "epoch": 0.825,
-      "grad_norm": 3.096435308456421,
-      "learning_rate": 3.7440000000000005e-06,
-      "loss": 0.1444,
+      "epoch": 0.66,
+      "grad_norm": 2.7979540824890137,
+      "learning_rate": 8.957894736842106e-06,
+      "loss": 0.1327,
       "step": 6600
     },
     {
-      "epoch": 0.828125,
-      "grad_norm": 3.5198802947998047,
-      "learning_rate": 3.6773333333333338e-06,
-      "loss": 0.1493,
+      "epoch": 0.6625,
+      "grad_norm": 3.4529471397399902,
+      "learning_rate": 8.892105263157895e-06,
+      "loss": 0.139,
       "step": 6625
     },
     {
-      "epoch": 0.83125,
-      "grad_norm": 3.3834660053253174,
-      "learning_rate": 3.6106666666666666e-06,
-      "loss": 0.1737,
+      "epoch": 0.665,
+      "grad_norm": 3.158491849899292,
+      "learning_rate": 8.826315789473685e-06,
+      "loss": 0.1613,
       "step": 6650
     },
     {
-      "epoch": 0.834375,
-      "grad_norm": 3.2613072395324707,
-      "learning_rate": 3.5440000000000003e-06,
-      "loss": 0.1408,
+      "epoch": 0.6675,
+      "grad_norm": 3.1579043865203857,
+      "learning_rate": 8.760526315789475e-06,
+      "loss": 0.1309,
       "step": 6675
     },
     {
-      "epoch": 0.8375,
-      "grad_norm": 2.618708848953247,
-      "learning_rate": 3.4773333333333336e-06,
-      "loss": 0.1427,
+      "epoch": 0.67,
+      "grad_norm": 2.4636425971984863,
+      "learning_rate": 8.694736842105264e-06,
+      "loss": 0.1327,
       "step": 6700
     },
     {
-      "epoch": 0.840625,
-      "grad_norm": 2.236100196838379,
-      "learning_rate": 3.4106666666666672e-06,
-      "loss": 0.0976,
+      "epoch": 0.6725,
+      "grad_norm": 1.9120688438415527,
+      "learning_rate": 8.628947368421053e-06,
+      "loss": 0.0921,
       "step": 6725
     },
     {
-      "epoch": 0.84375,
-      "grad_norm": 2.4650626182556152,
-      "learning_rate": 3.344e-06,
-      "loss": 0.0899,
+      "epoch": 0.675,
+      "grad_norm": 2.430513858795166,
+      "learning_rate": 8.563157894736842e-06,
+      "loss": 0.0828,
       "step": 6750
     },
     {
-      "epoch": 0.846875,
-      "grad_norm": 3.514897346496582,
-      "learning_rate": 3.2773333333333334e-06,
-      "loss": 0.1099,
+      "epoch": 0.6775,
+      "grad_norm": 3.101741075515747,
+      "learning_rate": 8.497368421052631e-06,
+      "loss": 0.102,
       "step": 6775
     },
     {
-      "epoch": 0.85,
-      "grad_norm": 3.703801155090332,
-      "learning_rate": 3.210666666666667e-06,
-      "loss": 0.1979,
+      "epoch": 0.68,
+      "grad_norm": 3.2060201168060303,
+      "learning_rate": 8.43157894736842e-06,
+      "loss": 0.1793,
       "step": 6800
     },
     {
-      "epoch": 0.853125,
-      "grad_norm": 4.976568698883057,
-      "learning_rate": 3.1440000000000003e-06,
-      "loss": 0.1801,
+      "epoch": 0.6825,
+      "grad_norm": 4.750592231750488,
+      "learning_rate": 8.365789473684211e-06,
+      "loss": 0.1581,
       "step": 6825
     },
     {
-      "epoch": 0.85625,
-      "grad_norm": 4.201725959777832,
-      "learning_rate": 3.077333333333334e-06,
-      "loss": 0.1839,
+      "epoch": 0.685,
+      "grad_norm": 3.6782851219177246,
+      "learning_rate": 8.3e-06,
+      "loss": 0.1657,
       "step": 6850
     },
     {
-      "epoch": 0.859375,
-      "grad_norm": 3.662229061126709,
-      "learning_rate": 3.010666666666667e-06,
-      "loss": 0.1695,
+      "epoch": 0.6875,
+      "grad_norm": 3.261650323867798,
+      "learning_rate": 8.23421052631579e-06,
+      "loss": 0.1562,
       "step": 6875
     },
     {
-      "epoch": 0.8625,
-      "grad_norm": 3.8069918155670166,
-      "learning_rate": 2.944e-06,
-      "loss": 0.1615,
+      "epoch": 0.69,
+      "grad_norm": 3.4499616622924805,
+      "learning_rate": 8.16842105263158e-06,
+      "loss": 0.1489,
       "step": 6900
     },
     {
-      "epoch": 0.865625,
-      "grad_norm": 3.208935499191284,
-      "learning_rate": 2.877333333333334e-06,
-      "loss": 0.1496,
+      "epoch": 0.6925,
+      "grad_norm": 3.001925230026245,
+      "learning_rate": 8.10263157894737e-06,
+      "loss": 0.1398,
       "step": 6925
     },
     {
-      "epoch": 0.86875,
-      "grad_norm": 3.4923043251037598,
-      "learning_rate": 2.810666666666667e-06,
-      "loss": 0.147,
+      "epoch": 0.695,
+      "grad_norm": 3.033891201019287,
+      "learning_rate": 8.036842105263158e-06,
+      "loss": 0.1354,
       "step": 6950
     },
     {
-      "epoch": 0.871875,
-      "grad_norm": 4.01771354675293,
-      "learning_rate": 2.744e-06,
-      "loss": 0.1751,
+      "epoch": 0.6975,
+      "grad_norm": 3.6877501010894775,
+      "learning_rate": 7.971052631578947e-06,
+      "loss": 0.1591,
       "step": 6975
     },
     {
-      "epoch": 0.875,
-      "grad_norm": 4.1294355392456055,
-      "learning_rate": 2.6773333333333336e-06,
-      "loss": 0.1733,
+      "epoch": 0.7,
+      "grad_norm": 3.818603754043579,
+      "learning_rate": 7.905263157894738e-06,
+      "loss": 0.1555,
       "step": 7000
     },
     {
-      "epoch": 0.875,
-      "eval_loss": 0.3004015386104584,
-      "eval_runtime": 74.1497,
-      "eval_samples_per_second": 28.375,
-      "eval_steps_per_second": 1.78,
-      "eval_wer": 13.933813218659438,
+      "epoch": 0.7,
+      "eval_loss": 0.37644582986831665,
+      "eval_runtime": 50.6902,
+      "eval_samples_per_second": 33.359,
+      "eval_steps_per_second": 2.091,
+      "eval_wer": 13.38204071437492,
       "step": 7000
     },
     {
-      "epoch": 0.878125,
-      "grad_norm": 12.625326156616211,
-      "learning_rate": 2.616e-06,
-      "loss": 0.3186,
+      "epoch": 0.7025,
+      "grad_norm": 6.232231616973877,
+      "learning_rate": 7.842105263157895e-06,
+      "loss": 0.2913,
       "step": 7025
     },
     {
-      "epoch": 0.88125,
-      "grad_norm": 7.057121753692627,
-      "learning_rate": 2.5493333333333337e-06,
-      "loss": 0.5086,
+      "epoch": 0.705,
+      "grad_norm": 7.377111434936523,
+      "learning_rate": 7.776315789473684e-06,
+      "loss": 0.4907,
       "step": 7050
     },
     {
-      "epoch": 0.884375,
-      "grad_norm": 5.440456390380859,
-      "learning_rate": 2.482666666666667e-06,
-      "loss": 0.481,
+      "epoch": 0.7075,
+      "grad_norm": 4.876298904418945,
+      "learning_rate": 7.710526315789473e-06,
+      "loss": 0.4457,
       "step": 7075
     },
     {
-      "epoch": 0.8875,
-      "grad_norm": 6.303742408752441,
-      "learning_rate": 2.4160000000000002e-06,
-      "loss": 0.4034,
+      "epoch": 0.71,
+      "grad_norm": 4.359467029571533,
+      "learning_rate": 7.644736842105264e-06,
+      "loss": 0.3599,
       "step": 7100
     },
     {
-      "epoch": 0.890625,
-      "grad_norm": 3.7720141410827637,
-      "learning_rate": 2.3493333333333335e-06,
-      "loss": 0.2274,
+      "epoch": 0.7125,
+      "grad_norm": 3.525312900543213,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.2052,
       "step": 7125
     },
     {
-      "epoch": 0.89375,
-      "grad_norm": 4.611368656158447,
-      "learning_rate": 2.2826666666666668e-06,
-      "loss": 0.1717,
+      "epoch": 0.715,
+      "grad_norm": 4.211317539215088,
+      "learning_rate": 7.513157894736843e-06,
+      "loss": 0.1551,
       "step": 7150
     },
     {
-      "epoch": 0.896875,
-      "grad_norm": 3.155137777328491,
-      "learning_rate": 2.216e-06,
-      "loss": 0.1594,
+      "epoch": 0.7175,
+      "grad_norm": 2.786363124847412,
+      "learning_rate": 7.447368421052632e-06,
+      "loss": 0.1428,
       "step": 7175
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 3.6036856174468994,
-      "learning_rate": 2.1493333333333337e-06,
-      "loss": 0.1264,
+      "epoch": 0.72,
+      "grad_norm": 2.948626756668091,
+      "learning_rate": 7.381578947368421e-06,
+      "loss": 0.1128,
       "step": 7200
     },
     {
-      "epoch": 0.903125,
-      "grad_norm": 3.040969133377075,
-      "learning_rate": 2.0826666666666666e-06,
-      "loss": 0.1198,
+      "epoch": 0.7225,
+      "grad_norm": 2.9625308513641357,
+      "learning_rate": 7.315789473684211e-06,
+      "loss": 0.1081,
       "step": 7225
     },
     {
-      "epoch": 0.90625,
-      "grad_norm": 2.538546562194824,
-      "learning_rate": 2.0160000000000003e-06,
-      "loss": 0.1233,
+      "epoch": 0.725,
+      "grad_norm": 2.201939821243286,
+      "learning_rate": 7.25e-06,
+      "loss": 0.1114,
       "step": 7250
     },
     {
-      "epoch": 0.909375,
-      "grad_norm": 2.2235965728759766,
-      "learning_rate": 1.9493333333333335e-06,
-      "loss": 0.0948,
+      "epoch": 0.7275,
+      "grad_norm": 2.0348243713378906,
+      "learning_rate": 7.184210526315791e-06,
+      "loss": 0.0882,
       "step": 7275
     },
     {
-      "epoch": 0.9125,
-      "grad_norm": 2.112567663192749,
-      "learning_rate": 1.8826666666666668e-06,
-      "loss": 0.0796,
+      "epoch": 0.73,
+      "grad_norm": 1.9231960773468018,
+      "learning_rate": 7.11842105263158e-06,
+      "loss": 0.0733,
       "step": 7300
     },
     {
-      "epoch": 0.915625,
-      "grad_norm": 2.5596227645874023,
-      "learning_rate": 1.8160000000000003e-06,
-      "loss": 0.0871,
+      "epoch": 0.7325,
+      "grad_norm": 2.2595067024230957,
+      "learning_rate": 7.052631578947369e-06,
+      "loss": 0.0793,
       "step": 7325
     },
     {
-      "epoch": 0.91875,
-      "grad_norm": 3.282794713973999,
-      "learning_rate": 1.7493333333333335e-06,
-      "loss": 0.1154,
+      "epoch": 0.735,
+      "grad_norm": 3.1157546043395996,
+      "learning_rate": 6.986842105263158e-06,
+      "loss": 0.106,
       "step": 7350
     },
     {
-      "epoch": 0.921875,
-      "grad_norm": 3.568565607070923,
-      "learning_rate": 1.6826666666666668e-06,
-      "loss": 0.1648,
+      "epoch": 0.7375,
+      "grad_norm": 2.9897453784942627,
+      "learning_rate": 6.921052631578947e-06,
+      "loss": 0.1448,
       "step": 7375
     },
     {
-      "epoch": 0.925,
-      "grad_norm": 3.731203079223633,
-      "learning_rate": 1.616e-06,
-      "loss": 0.132,
+      "epoch": 0.74,
+      "grad_norm": 3.1483774185180664,
+      "learning_rate": 6.8552631578947365e-06,
+      "loss": 0.1132,
       "step": 7400
     },
     {
-      "epoch": 0.928125,
-      "grad_norm": 2.649831771850586,
-      "learning_rate": 1.5493333333333335e-06,
-      "loss": 0.1518,
+      "epoch": 0.7425,
+      "grad_norm": 2.285529136657715,
+      "learning_rate": 6.7894736842105264e-06,
+      "loss": 0.1292,
       "step": 7425
     },
     {
-      "epoch": 0.93125,
-      "grad_norm": 2.2203938961029053,
-      "learning_rate": 1.4826666666666666e-06,
-      "loss": 0.1045,
+      "epoch": 0.745,
+      "grad_norm": 2.1281116008758545,
+      "learning_rate": 6.723684210526316e-06,
+      "loss": 0.0942,
       "step": 7450
     },
     {
-      "epoch": 0.934375,
-      "grad_norm": 3.9395177364349365,
-      "learning_rate": 1.416e-06,
-      "loss": 0.0957,
+      "epoch": 0.7475,
+      "grad_norm": 2.2459075450897217,
+      "learning_rate": 6.657894736842106e-06,
+      "loss": 0.0887,
       "step": 7475
     },
     {
-      "epoch": 0.9375,
-      "grad_norm": 3.2605130672454834,
-      "learning_rate": 1.3493333333333333e-06,
-      "loss": 0.0901,
+      "epoch": 0.75,
+      "grad_norm": 2.9605884552001953,
+      "learning_rate": 6.5921052631578955e-06,
+      "loss": 0.0824,
       "step": 7500
     },
     {
-      "epoch": 0.940625,
-      "grad_norm": 4.289961814880371,
-      "learning_rate": 1.2826666666666668e-06,
-      "loss": 0.1846,
+      "epoch": 0.7525,
+      "grad_norm": 3.7873899936676025,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.1646,
       "step": 7525
     },
     {
-      "epoch": 0.94375,
-      "grad_norm": 4.553671836853027,
-      "learning_rate": 1.216e-06,
-      "loss": 0.2492,
+      "epoch": 0.755,
+      "grad_norm": 4.234455585479736,
+      "learning_rate": 6.460526315789474e-06,
+      "loss": 0.2172,
       "step": 7550
     },
     {
-      "epoch": 0.946875,
-      "grad_norm": 5.279869079589844,
-      "learning_rate": 1.1493333333333334e-06,
-      "loss": 0.2844,
+      "epoch": 0.7575,
+      "grad_norm": 4.279770374298096,
+      "learning_rate": 6.394736842105263e-06,
+      "loss": 0.2445,
       "step": 7575
     },
     {
-      "epoch": 0.95,
-      "grad_norm": 2.1614372730255127,
-      "learning_rate": 1.0826666666666668e-06,
-      "loss": 0.1715,
+      "epoch": 0.76,
+      "grad_norm": 1.964227557182312,
+      "learning_rate": 6.328947368421052e-06,
+      "loss": 0.1534,
       "step": 7600
     },
     {
-      "epoch": 0.953125,
-      "grad_norm": 2.8423452377319336,
-      "learning_rate": 1.016e-06,
-      "loss": 0.1162,
+      "epoch": 0.7625,
+      "grad_norm": 2.5846900939941406,
+      "learning_rate": 6.263157894736843e-06,
+      "loss": 0.1069,
       "step": 7625
     },
     {
-      "epoch": 0.95625,
-      "grad_norm": 2.077169895172119,
-      "learning_rate": 9.493333333333334e-07,
-      "loss": 0.094,
+      "epoch": 0.765,
+      "grad_norm": 2.0265769958496094,
+      "learning_rate": 6.197368421052632e-06,
+      "loss": 0.0853,
       "step": 7650
     },
     {
-      "epoch": 0.959375,
-      "grad_norm": 3.680450201034546,
-      "learning_rate": 8.826666666666666e-07,
-      "loss": 0.1291,
+      "epoch": 0.7675,
+      "grad_norm": 4.039712905883789,
+      "learning_rate": 6.131578947368422e-06,
+      "loss": 0.1171,
       "step": 7675
     },
     {
-      "epoch": 0.9625,
-      "grad_norm": 3.4633026123046875,
-      "learning_rate": 8.160000000000001e-07,
-      "loss": 0.1549,
+      "epoch": 0.77,
+      "grad_norm": 5.171535015106201,
+      "learning_rate": 6.065789473684211e-06,
+      "loss": 0.1451,
       "step": 7700
     },
     {
-      "epoch": 0.965625,
-      "grad_norm": 3.8427698612213135,
-      "learning_rate": 7.493333333333335e-07,
-      "loss": 0.1646,
+      "epoch": 0.7725,
+      "grad_norm": 3.3763575553894043,
+      "learning_rate": 6e-06,
+      "loss": 0.1522,
       "step": 7725
     },
     {
-      "epoch": 0.96875,
-      "grad_norm": 2.8945538997650146,
-      "learning_rate": 6.826666666666667e-07,
-      "loss": 0.1524,
+      "epoch": 0.775,
+      "grad_norm": 2.777357578277588,
+      "learning_rate": 5.93421052631579e-06,
+      "loss": 0.1395,
       "step": 7750
     },
     {
-      "epoch": 0.971875,
-      "grad_norm": 3.424391269683838,
-      "learning_rate": 6.160000000000001e-07,
-      "loss": 0.1083,
+      "epoch": 0.7775,
+      "grad_norm": 3.0905845165252686,
+      "learning_rate": 5.868421052631579e-06,
+      "loss": 0.098,
       "step": 7775
     },
     {
-      "epoch": 0.975,
-      "grad_norm": 5.2906951904296875,
-      "learning_rate": 5.493333333333334e-07,
-      "loss": 0.1327,
+      "epoch": 0.78,
+      "grad_norm": 4.638239860534668,
+      "learning_rate": 5.802631578947368e-06,
+      "loss": 0.1158,
       "step": 7800
     },
     {
-      "epoch": 0.978125,
-      "grad_norm": 6.5452046394348145,
-      "learning_rate": 4.853333333333333e-07,
-      "loss": 0.3624,
+      "epoch": 0.7825,
+      "grad_norm": 3.2748498916625977,
+      "learning_rate": 5.739473684210526e-06,
+      "loss": 0.2657,
       "step": 7825
     },
     {
-      "epoch": 0.98125,
-      "grad_norm": 2.549628496170044,
-      "learning_rate": 4.186666666666667e-07,
-      "loss": 0.1518,
+      "epoch": 0.785,
+      "grad_norm": 2.237273931503296,
+      "learning_rate": 5.6736842105263155e-06,
+      "loss": 0.1286,
       "step": 7850
     },
     {
-      "epoch": 0.984375,
-      "grad_norm": 2.076683759689331,
-      "learning_rate": 3.5200000000000003e-07,
-      "loss": 0.1048,
+      "epoch": 0.7875,
+      "grad_norm": 1.864817500114441,
+      "learning_rate": 5.6078947368421054e-06,
+      "loss": 0.0943,
       "step": 7875
     },
     {
-      "epoch": 0.9875,
-      "grad_norm": 2.8122971057891846,
-      "learning_rate": 2.8533333333333335e-07,
-      "loss": 0.0981,
+      "epoch": 0.79,
+      "grad_norm": 2.4019246101379395,
+      "learning_rate": 5.542105263157895e-06,
+      "loss": 0.0889,
       "step": 7900
     },
     {
-      "epoch": 0.990625,
-      "grad_norm": 2.3224234580993652,
-      "learning_rate": 2.186666666666667e-07,
-      "loss": 0.1017,
+      "epoch": 0.7925,
+      "grad_norm": 2.1701340675354004,
+      "learning_rate": 5.4763157894736845e-06,
+      "loss": 0.093,
       "step": 7925
     },
     {
-      "epoch": 0.99375,
-      "grad_norm": 2.10176682472229,
-      "learning_rate": 1.52e-07,
-      "loss": 0.0937,
+      "epoch": 0.795,
+      "grad_norm": 2.0279428958892822,
+      "learning_rate": 5.4105263157894745e-06,
+      "loss": 0.0839,
       "step": 7950
     },
     {
-      "epoch": 0.996875,
-      "grad_norm": 3.3718252182006836,
-      "learning_rate": 8.533333333333334e-08,
-      "loss": 0.1,
+      "epoch": 0.7975,
+      "grad_norm": 2.8084793090820312,
+      "learning_rate": 5.344736842105264e-06,
+      "loss": 0.0906,
       "step": 7975
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 4.459543704986572,
-      "learning_rate": 1.866666666666667e-08,
-      "loss": 0.1228,
+      "epoch": 0.8,
+      "grad_norm": 2.640064001083374,
+      "learning_rate": 5.278947368421053e-06,
+      "loss": 0.114,
       "step": 8000
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 0.24521400034427643,
-      "eval_runtime": 75.7252,
-      "eval_samples_per_second": 27.785,
-      "eval_steps_per_second": 1.743,
-      "eval_wer": 13.816958025614658,
+      "epoch": 0.8,
+      "eval_loss": 0.3096904158592224,
+      "eval_runtime": 50.4876,
+      "eval_samples_per_second": 33.493,
+      "eval_steps_per_second": 2.1,
+      "eval_wer": 12.75134257524666,
       "step": 8000
     },
+    {
+      "epoch": 0.8025,
+      "grad_norm": 3.118774652481079,
+      "learning_rate": 5.213157894736842e-06,
+      "loss": 0.1362,
+      "step": 8025
+    },
+    {
+      "epoch": 0.805,
+      "grad_norm": 3.070131778717041,
+      "learning_rate": 5.147368421052632e-06,
+      "loss": 0.1416,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8075,
+      "grad_norm": 2.8992044925689697,
+      "learning_rate": 5.081578947368421e-06,
+      "loss": 0.1097,
+      "step": 8075
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2826101779937744,
+      "learning_rate": 5.015789473684211e-06,
+      "loss": 0.0972,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8125,
+      "grad_norm": 2.396894693374634,
+      "learning_rate": 4.950000000000001e-06,
+      "loss": 0.0823,
+      "step": 8125
+    },
+    {
+      "epoch": 0.815,
+      "grad_norm": 3.337656259536743,
+      "learning_rate": 4.88421052631579e-06,
+      "loss": 0.0834,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8175,
+      "grad_norm": 1.498647689819336,
+      "learning_rate": 4.818421052631579e-06,
+      "loss": 0.0732,
+      "step": 8175
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0886967182159424,
+      "learning_rate": 4.752631578947369e-06,
+      "loss": 0.0632,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8225,
+      "grad_norm": 1.7843704223632812,
+      "learning_rate": 4.686842105263158e-06,
+      "loss": 0.0767,
+      "step": 8225
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 2.3446483612060547,
+      "learning_rate": 4.621052631578947e-06,
+      "loss": 0.0718,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8275,
+      "grad_norm": 2.0303142070770264,
+      "learning_rate": 4.5552631578947365e-06,
+      "loss": 0.0733,
+      "step": 8275
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4907219409942627,
+      "learning_rate": 4.4894736842105264e-06,
+      "loss": 0.0751,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8325,
+      "grad_norm": 3.2890570163726807,
+      "learning_rate": 4.4236842105263156e-06,
+      "loss": 0.1089,
+      "step": 8325
+    },
+    {
+      "epoch": 0.835,
+      "grad_norm": 3.4118282794952393,
+      "learning_rate": 4.3578947368421055e-06,
+      "loss": 0.1413,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8375,
+      "grad_norm": 4.141169548034668,
+      "learning_rate": 4.2921052631578955e-06,
+      "loss": 0.1615,
+      "step": 8375
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5593221187591553,
+      "learning_rate": 4.226315789473685e-06,
+      "loss": 0.1325,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8425,
+      "grad_norm": 2.8442673683166504,
+      "learning_rate": 4.160526315789474e-06,
+      "loss": 0.0894,
+      "step": 8425
+    },
+    {
+      "epoch": 0.845,
+      "grad_norm": 2.300579786300659,
+      "learning_rate": 4.094736842105263e-06,
+      "loss": 0.0857,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8475,
+      "grad_norm": 2.8144922256469727,
+      "learning_rate": 4.028947368421053e-06,
+      "loss": 0.0977,
+      "step": 8475
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6928155422210693,
+      "learning_rate": 3.963157894736842e-06,
+      "loss": 0.1106,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8525,
+      "grad_norm": 2.9384024143218994,
+      "learning_rate": 3.897368421052631e-06,
+      "loss": 0.1139,
+      "step": 8525
+    },
+    {
+      "epoch": 0.855,
+      "grad_norm": 4.728883266448975,
+      "learning_rate": 3.831578947368421e-06,
+      "loss": 0.1597,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8575,
+      "grad_norm": 4.498551845550537,
+      "learning_rate": 3.7657894736842106e-06,
+      "loss": 0.2747,
+      "step": 8575
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.742629528045654,
+      "learning_rate": 3.7e-06,
+      "loss": 0.3854,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8625,
+      "grad_norm": 6.729153633117676,
+      "learning_rate": 3.63421052631579e-06,
+      "loss": 0.456,
+      "step": 8625
+    },
+    {
+      "epoch": 0.865,
+      "grad_norm": 2.6989076137542725,
+      "learning_rate": 3.5684210526315792e-06,
+      "loss": 0.2648,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8675,
+      "grad_norm": 4.221801280975342,
+      "learning_rate": 3.5026315789473684e-06,
+      "loss": 0.1734,
+      "step": 8675
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 9.831379890441895,
+      "learning_rate": 3.436842105263158e-06,
+      "loss": 0.1989,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8725,
+      "grad_norm": 2.7033917903900146,
+      "learning_rate": 3.371052631578948e-06,
+      "loss": 0.1374,
+      "step": 8725
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 2.1275370121002197,
+      "learning_rate": 3.305263157894737e-06,
+      "loss": 0.0852,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8775,
+      "grad_norm": 2.1816318035125732,
+      "learning_rate": 3.239473684210526e-06,
+      "loss": 0.0907,
+      "step": 8775
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.410520553588867,
+      "learning_rate": 3.173684210526316e-06,
+      "loss": 0.1015,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8825,
+      "grad_norm": 3.198993444442749,
+      "learning_rate": 3.1078947368421056e-06,
+      "loss": 0.1253,
+      "step": 8825
+    },
+    {
+      "epoch": 0.885,
+      "grad_norm": 3.826889753341675,
+      "learning_rate": 3.0421052631578948e-06,
+      "loss": 0.123,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8875,
+      "grad_norm": 2.976738929748535,
+      "learning_rate": 2.9763157894736843e-06,
+      "loss": 0.1254,
+      "step": 8875
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.686645746231079,
+      "learning_rate": 2.910526315789474e-06,
+      "loss": 0.0858,
+      "step": 8900
+    },
+    {
+      "epoch": 0.8925,
+      "grad_norm": 2.974180221557617,
+      "learning_rate": 2.8447368421052634e-06,
+      "loss": 0.0841,
+      "step": 8925
+    },
+    {
+      "epoch": 0.895,
+      "grad_norm": 2.810300827026367,
+      "learning_rate": 2.778947368421053e-06,
+      "loss": 0.0906,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8975,
+      "grad_norm": 2.3508145809173584,
+      "learning_rate": 2.713157894736842e-06,
+      "loss": 0.0739,
+      "step": 8975
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.418074369430542,
+      "learning_rate": 2.6473684210526316e-06,
+      "loss": 0.0775,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.31704333424568176,
+      "eval_runtime": 50.8782,
+      "eval_samples_per_second": 33.236,
+      "eval_steps_per_second": 2.083,
+      "eval_wer": 12.457849381790934,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9025,
+      "grad_norm": 2.4813151359558105,
+      "learning_rate": 2.581578947368421e-06,
+      "loss": 0.0898,
+      "step": 9025
+    },
+    {
+      "epoch": 0.905,
+      "grad_norm": 3.3182260990142822,
+      "learning_rate": 2.5157894736842107e-06,
+      "loss": 0.1079,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9075,
+      "grad_norm": 4.108717441558838,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.1409,
+      "step": 9075
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0588371753692627,
+      "learning_rate": 2.3842105263157894e-06,
+      "loss": 0.1391,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9125,
+      "grad_norm": 2.8071908950805664,
+      "learning_rate": 2.318421052631579e-06,
+      "loss": 0.1108,
+      "step": 9125
+    },
+    {
+      "epoch": 0.915,
+      "grad_norm": 2.184046506881714,
+      "learning_rate": 2.2526315789473685e-06,
+      "loss": 0.1049,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9175,
+      "grad_norm": 2.5016894340515137,
+      "learning_rate": 2.186842105263158e-06,
+      "loss": 0.1009,
+      "step": 9175
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4719274044036865,
+      "learning_rate": 2.1210526315789476e-06,
+      "loss": 0.105,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9225,
+      "grad_norm": 3.3513808250427246,
+      "learning_rate": 2.0552631578947367e-06,
+      "loss": 0.1261,
+      "step": 9225
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 4.014399528503418,
+      "learning_rate": 1.9894736842105267e-06,
+      "loss": 0.138,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9275,
+      "grad_norm": 2.7651119232177734,
+      "learning_rate": 1.9236842105263158e-06,
+      "loss": 0.1542,
+      "step": 9275
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8056927919387817,
+      "learning_rate": 1.8578947368421053e-06,
+      "loss": 0.096,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9325,
+      "grad_norm": 2.349053144454956,
+      "learning_rate": 1.7921052631578949e-06,
+      "loss": 0.0983,
+      "step": 9325
+    },
+    {
+      "epoch": 0.935,
+      "grad_norm": 1.9038828611373901,
+      "learning_rate": 1.7263157894736842e-06,
+      "loss": 0.0809,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 2.875915288925171,
+      "learning_rate": 1.6605263157894738e-06,
+      "loss": 0.0976,
+      "step": 9375
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.029100179672241,
+      "learning_rate": 1.594736842105263e-06,
+      "loss": 0.1235,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9425,
+      "grad_norm": 2.9959537982940674,
+      "learning_rate": 1.5289473684210526e-06,
+      "loss": 0.1426,
+      "step": 9425
+    },
+    {
+      "epoch": 0.945,
+      "grad_norm": 2.2764275074005127,
+      "learning_rate": 1.4631578947368422e-06,
+      "loss": 0.2688,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9475,
+      "grad_norm": 2.720301389694214,
+      "learning_rate": 1.3973684210526315e-06,
+      "loss": 0.1352,
+      "step": 9475
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.853646993637085,
+      "learning_rate": 1.331578947368421e-06,
+      "loss": 0.1102,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9525,
+      "grad_norm": 3.4527268409729004,
+      "learning_rate": 1.2657894736842106e-06,
+      "loss": 0.1336,
+      "step": 9525
+    },
+    {
+      "epoch": 0.955,
+      "grad_norm": 3.289494514465332,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.141,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9575,
+      "grad_norm": 3.2890825271606445,
+      "learning_rate": 1.1342105263157895e-06,
+      "loss": 0.1591,
+      "step": 9575
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4674901962280273,
+      "learning_rate": 1.068421052631579e-06,
+      "loss": 0.123,
+      "step": 9600
+    },
+    {
+      "epoch": 0.9625,
+      "grad_norm": 2.3272643089294434,
+      "learning_rate": 1.0026315789473684e-06,
+      "loss": 0.0979,
+      "step": 9625
+    },
+    {
+      "epoch": 0.965,
+      "grad_norm": 2.303502321243286,
+      "learning_rate": 9.36842105263158e-07,
+      "loss": 0.0761,
+      "step": 9650
+    },
+    {
+      "epoch": 0.9675,
+      "grad_norm": 2.0139310359954834,
+      "learning_rate": 8.710526315789475e-07,
+      "loss": 0.0765,
+      "step": 9675
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.459679126739502,
+      "learning_rate": 8.052631578947369e-07,
+      "loss": 0.078,
+      "step": 9700
+    },
+    {
+      "epoch": 0.9725,
+      "grad_norm": 2.502119541168213,
+      "learning_rate": 7.394736842105263e-07,
+      "loss": 0.0764,
+      "step": 9725
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 3.774677038192749,
+      "learning_rate": 6.736842105263159e-07,
+      "loss": 0.0822,
+      "step": 9750
+    },
+    {
+      "epoch": 0.9775,
+      "grad_norm": 2.353592872619629,
+      "learning_rate": 6.078947368421053e-07,
+      "loss": 0.093,
+      "step": 9775
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.306469202041626,
+      "learning_rate": 5.421052631578948e-07,
+      "loss": 0.1237,
+      "step": 9800
+    },
+    {
+      "epoch": 0.9825,
+      "grad_norm": 3.7858200073242188,
+      "learning_rate": 4.7631578947368427e-07,
+      "loss": 0.1381,
+      "step": 9825
+    },
+    {
+      "epoch": 0.985,
+      "grad_norm": 2.6253809928894043,
+      "learning_rate": 4.105263157894737e-07,
+      "loss": 0.117,
+      "step": 9850
+    },
+    {
+      "epoch": 0.9875,
+      "grad_norm": 2.679793119430542,
+      "learning_rate": 3.447368421052632e-07,
+      "loss": 0.1253,
+      "step": 9875
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.4605584144592285,
+      "learning_rate": 2.7894736842105264e-07,
+      "loss": 0.1865,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9925,
+      "grad_norm": 1.9926120042800903,
+      "learning_rate": 2.131578947368421e-07,
+      "loss": 0.1615,
+      "step": 9925
+    },
+    {
+      "epoch": 0.995,
+      "grad_norm": 2.260565757751465,
+      "learning_rate": 1.473684210526316e-07,
+      "loss": 0.0843,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9975,
+      "grad_norm": 2.2065985202789307,
+      "learning_rate": 8.157894736842106e-08,
+      "loss": 0.0841,
+      "step": 9975
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.219501256942749,
+      "learning_rate": 1.578947368421053e-08,
+      "loss": 0.0836,
+      "step": 10000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.3182850182056427,
+      "eval_runtime": 51.7745,
+      "eval_samples_per_second": 32.661,
+      "eval_steps_per_second": 2.047,
+      "eval_wer": 12.307980517047582,
+      "step": 10000
+    },
     {
       "epoch": 1.0,
-      "step": 8000,
-      "total_flos": 1.660415901696e+19,
-      "train_loss": 0.22206098145246506,
-      "train_runtime": 4270.5513,
-      "train_samples_per_second": 59.945,
-      "train_steps_per_second": 1.873
+      "step": 10000,
+      "total_flos": 2.07551987712e+19,
+      "train_loss": 0.1926619317293167,
+      "train_runtime": 5040.4481,
+      "train_samples_per_second": 63.486,
+      "train_steps_per_second": 1.984
     }
   ],
   "logging_steps": 25,
-  "max_steps": 8000,
+  "max_steps": 10000,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 9223372036854775807,
   "save_steps": 1000,
@@ -2347,7 +2925,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.660415901696e+19,
+  "total_flos": 2.07551987712e+19,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null