diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5842 @@
+{
+  "best_metric": 90.76850542197077,
+  "best_model_checkpoint": "./iteboshi_student_model_temp/checkpoint-5000",
+  "epoch": 22.026431718061673,
+  "eval_steps": 1000,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02753303964757709,
+      "grad_norm": 7.0630106925964355,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 11.955,
+      "step": 25
+    },
+    {
+      "epoch": 0.05506607929515418,
+      "grad_norm": 5.473441123962402,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 10.7888,
+      "step": 50
+    },
+    {
+      "epoch": 0.08259911894273128,
+      "grad_norm": 3.244906187057495,
+      "learning_rate": 3e-06,
+      "loss": 8.3248,
+      "step": 75
+    },
+    {
+      "epoch": 0.11013215859030837,
+      "grad_norm": 2.3983211517333984,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 6.2859,
+      "step": 100
+    },
+    {
+      "epoch": 0.13766519823788545,
+      "grad_norm": 1.7205311059951782,
+      "learning_rate": 5e-06,
+      "loss": 5.1612,
+      "step": 125
+    },
+    {
+      "epoch": 0.16519823788546256,
+      "grad_norm": 1.4956705570220947,
+      "learning_rate": 6e-06,
+      "loss": 4.6601,
+      "step": 150
+    },
+    {
+      "epoch": 0.19273127753303965,
+      "grad_norm": 1.3801974058151245,
+      "learning_rate": 7e-06,
+      "loss": 4.2995,
+      "step": 175
+    },
+    {
+      "epoch": 0.22026431718061673,
+      "grad_norm": 1.2982876300811768,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.1134,
+      "step": 200
+    },
+    {
+      "epoch": 0.24779735682819384,
+      "grad_norm": 4.815393447875977,
+      "learning_rate": 9e-06,
+      "loss": 3.9592,
+      "step": 225
+    },
+    {
+      "epoch": 0.2753303964757709,
+      "grad_norm": 1.2964017391204834,
+      "learning_rate": 1e-05,
+      "loss": 3.8346,
+      "step": 250
+    },
+    {
+      "epoch": 0.30286343612334804,
+      "grad_norm": 1.2769396305084229,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 3.7025,
+      "step": 275
+    },
+    {
+      "epoch": 0.3303964757709251,
+      "grad_norm": 1.2632490396499634,
+      "learning_rate": 1.2e-05,
+      "loss": 3.5762,
+      "step": 300
+    },
+    {
+      "epoch": 0.3579295154185022,
+      "grad_norm": 1.296427607536316,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 3.4899,
+      "step": 325
+    },
+    {
+      "epoch": 0.3854625550660793,
+      "grad_norm": 1.271353006362915,
+      "learning_rate": 1.4e-05,
+      "loss": 3.4158,
+      "step": 350
+    },
+    {
+      "epoch": 0.4129955947136564,
+      "grad_norm": 1.4885754585266113,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 3.2939,
+      "step": 375
+    },
+    {
+      "epoch": 0.44052863436123346,
+      "grad_norm": 1.4233094453811646,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 3.1291,
+      "step": 400
+    },
+    {
+      "epoch": 0.46806167400881055,
+      "grad_norm": 1.2837520837783813,
+      "learning_rate": 1.7e-05,
+      "loss": 3.0305,
+      "step": 425
+    },
+    {
+      "epoch": 0.4955947136563877,
+      "grad_norm": 1.5688410997390747,
+      "learning_rate": 1.8e-05,
+      "loss": 2.9491,
+      "step": 450
+    },
+    {
+      "epoch": 0.5231277533039648,
+      "grad_norm": 1.5300930738449097,
+      "learning_rate": 1.9e-05,
+      "loss": 2.8143,
+      "step": 475
+    },
+    {
+      "epoch": 0.5506607929515418,
+      "grad_norm": 1.5693153142929077,
+      "learning_rate": 2e-05,
+      "loss": 2.6308,
+      "step": 500
+    },
+    {
+      "epoch": 0.5781938325991189,
+      "grad_norm": 1.5616507530212402,
+      "learning_rate": 1.9974358974358975e-05,
+      "loss": 2.5482,
+      "step": 525
+    },
+    {
+      "epoch": 0.6057268722466961,
+      "grad_norm": 1.5664052963256836,
+      "learning_rate": 1.994871794871795e-05,
+      "loss": 2.3664,
+      "step": 550
+    },
+    {
+      "epoch": 0.6332599118942731,
+      "grad_norm": 1.5610151290893555,
+      "learning_rate": 1.9923076923076926e-05,
+      "loss": 2.2403,
+      "step": 575
+    },
+    {
+      "epoch": 0.6607929515418502,
+      "grad_norm": 1.4449105262756348,
+      "learning_rate": 1.98974358974359e-05,
+      "loss": 2.0901,
+      "step": 600
+    },
+    {
+      "epoch": 0.6883259911894273,
+      "grad_norm": 1.5401190519332886,
+      "learning_rate": 1.9871794871794873e-05,
+      "loss": 1.9593,
+      "step": 625
+    },
+    {
+      "epoch": 0.7158590308370044,
+      "grad_norm": 1.6314514875411987,
+      "learning_rate": 1.9846153846153847e-05,
+      "loss": 1.9258,
+      "step": 650
+    },
+    {
+      "epoch": 0.7433920704845814,
+      "grad_norm": 2.754133939743042,
+      "learning_rate": 1.9820512820512824e-05,
+      "loss": 1.8044,
+      "step": 675
+    },
+    {
+      "epoch": 0.7709251101321586,
+      "grad_norm": 1.4502687454223633,
+      "learning_rate": 1.9794871794871798e-05,
+      "loss": 1.7527,
+      "step": 700
+    },
+    {
+      "epoch": 0.7984581497797357,
+      "grad_norm": 2.3516573905944824,
+      "learning_rate": 1.976923076923077e-05,
+      "loss": 1.7341,
+      "step": 725
+    },
+    {
+      "epoch": 0.8259911894273128,
+      "grad_norm": 1.7115260362625122,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.6904,
+      "step": 750
+    },
+    {
+      "epoch": 0.8535242290748899,
+      "grad_norm": 2.4372172355651855,
+      "learning_rate": 1.9717948717948722e-05,
+      "loss": 1.562,
+      "step": 775
+    },
+    {
+      "epoch": 0.8810572687224669,
+      "grad_norm": 1.575111746788025,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 1.556,
+      "step": 800
+    },
+    {
+      "epoch": 0.9085903083700441,
+      "grad_norm": 2.196908473968506,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 1.4173,
+      "step": 825
+    },
+    {
+      "epoch": 0.9361233480176211,
+      "grad_norm": 2.892843008041382,
+      "learning_rate": 1.9641025641025643e-05,
+      "loss": 1.453,
+      "step": 850
+    },
+    {
+      "epoch": 0.9636563876651982,
+      "grad_norm": 1.197956919670105,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 1.4298,
+      "step": 875
+    },
+    {
+      "epoch": 0.9911894273127754,
+      "grad_norm": 3.7183637619018555,
+      "learning_rate": 1.958974358974359e-05,
+      "loss": 1.3977,
+      "step": 900
+    },
+    {
+      "epoch": 1.0187224669603525,
+      "grad_norm": 3.0768394470214844,
+      "learning_rate": 1.9564102564102564e-05,
+      "loss": 1.3247,
+      "step": 925
+    },
+    {
+      "epoch": 1.0462555066079295,
+      "grad_norm": 1.5889824628829956,
+      "learning_rate": 1.953846153846154e-05,
+      "loss": 1.1798,
+      "step": 950
+    },
+    {
+      "epoch": 1.0737885462555066,
+      "grad_norm": 1.464847207069397,
+      "learning_rate": 1.9512820512820515e-05,
+      "loss": 1.1679,
+      "step": 975
+    },
+    {
+      "epoch": 1.1013215859030836,
+      "grad_norm": 1.1854747533798218,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.1451,
+      "step": 1000
+    },
+    {
+      "epoch": 1.1013215859030836,
+      "eval_cer": 52.99542604289661,
+      "eval_loss": 1.312658667564392,
+      "eval_runtime": 802.8816,
+      "eval_samples_per_second": 13.179,
+      "eval_steps_per_second": 3.296,
+      "eval_wer": 98.00094295143799,
+      "step": 1000
+    },
+    {
+      "epoch": 1.1288546255506609,
+      "grad_norm": 1.2571742534637451,
+      "learning_rate": 1.9461538461538462e-05,
+      "loss": 1.138,
+      "step": 1025
+    },
+    {
+      "epoch": 1.1563876651982379,
+      "grad_norm": 2.522019624710083,
+      "learning_rate": 1.943589743589744e-05,
+      "loss": 1.1544,
+      "step": 1050
+    },
+    {
+      "epoch": 1.183920704845815,
+      "grad_norm": 2.5862576961517334,
+      "learning_rate": 1.9410256410256413e-05,
+      "loss": 1.1321,
+      "step": 1075
+    },
+    {
+      "epoch": 1.2114537444933922,
+      "grad_norm": 1.393262267112732,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 1.063,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2389867841409692,
+      "grad_norm": 1.243342638015747,
+      "learning_rate": 1.935897435897436e-05,
+      "loss": 1.137,
+      "step": 1125
+    },
+    {
+      "epoch": 1.2665198237885462,
+      "grad_norm": 1.507082462310791,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 1.053,
+      "step": 1150
+    },
+    {
+      "epoch": 1.2940528634361232,
+      "grad_norm": 1.4709209203720093,
+      "learning_rate": 1.930769230769231e-05,
+      "loss": 1.0388,
+      "step": 1175
+    },
+    {
+      "epoch": 1.3215859030837005,
+      "grad_norm": 1.3831913471221924,
+      "learning_rate": 1.9282051282051284e-05,
+      "loss": 1.075,
+      "step": 1200
+    },
+    {
+      "epoch": 1.3491189427312775,
+      "grad_norm": 2.1179986000061035,
+      "learning_rate": 1.9256410256410258e-05,
+      "loss": 1.0266,
+      "step": 1225
+    },
+    {
+      "epoch": 1.3766519823788546,
+      "grad_norm": 2.009942054748535,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.0482,
+      "step": 1250
+    },
+    {
+      "epoch": 1.4041850220264318,
+      "grad_norm": 1.2496631145477295,
+      "learning_rate": 1.920512820512821e-05,
+      "loss": 0.9894,
+      "step": 1275
+    },
+    {
+      "epoch": 1.4317180616740088,
+      "grad_norm": 1.1840794086456299,
+      "learning_rate": 1.9179487179487182e-05,
+      "loss": 0.9793,
+      "step": 1300
+    },
+    {
+      "epoch": 1.4592511013215859,
+      "grad_norm": 1.4337009191513062,
+      "learning_rate": 1.9153846153846156e-05,
+      "loss": 1.037,
+      "step": 1325
+    },
+    {
+      "epoch": 1.4867841409691631,
+      "grad_norm": 1.1954481601715088,
+      "learning_rate": 1.912820512820513e-05,
+      "loss": 0.9891,
+      "step": 1350
+    },
+    {
+      "epoch": 1.51431718061674,
+      "grad_norm": 1.710187554359436,
+      "learning_rate": 1.9102564102564106e-05,
+      "loss": 0.9872,
+      "step": 1375
+    },
+    {
+      "epoch": 1.5418502202643172,
+      "grad_norm": 1.7128686904907227,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.996,
+      "step": 1400
+    },
+    {
+      "epoch": 1.5693832599118944,
+      "grad_norm": 1.127489686012268,
+      "learning_rate": 1.905128205128205e-05,
+      "loss": 0.9847,
+      "step": 1425
+    },
+    {
+      "epoch": 1.5969162995594712,
+      "grad_norm": 2.570929765701294,
+      "learning_rate": 1.9025641025641027e-05,
+      "loss": 0.9708,
+      "step": 1450
+    },
+    {
+      "epoch": 1.6244493392070485,
+      "grad_norm": 1.8463544845581055,
+      "learning_rate": 1.9e-05,
+      "loss": 0.9262,
+      "step": 1475
+    },
+    {
+      "epoch": 1.6519823788546255,
+      "grad_norm": 2.259000062942505,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.8969,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6795154185022025,
+      "grad_norm": 1.1015965938568115,
+      "learning_rate": 1.894871794871795e-05,
+      "loss": 0.9137,
+      "step": 1525
+    },
+    {
+      "epoch": 1.7070484581497798,
+      "grad_norm": 1.5946825742721558,
+      "learning_rate": 1.8923076923076925e-05,
+      "loss": 0.9043,
+      "step": 1550
+    },
+    {
+      "epoch": 1.7345814977973568,
+      "grad_norm": 1.1002482175827026,
+      "learning_rate": 1.88974358974359e-05,
+      "loss": 0.8996,
+      "step": 1575
+    },
+    {
+      "epoch": 1.7621145374449338,
+      "grad_norm": 1.141486644744873,
+      "learning_rate": 1.8871794871794873e-05,
+      "loss": 0.9047,
+      "step": 1600
+    },
+    {
+      "epoch": 1.789647577092511,
+      "grad_norm": 1.5526583194732666,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.957,
+      "step": 1625
+    },
+    {
+      "epoch": 1.8171806167400881,
+      "grad_norm": 2.08101224899292,
+      "learning_rate": 1.8820512820512823e-05,
+      "loss": 0.9406,
+      "step": 1650
+    },
+    {
+      "epoch": 1.8447136563876652,
+      "grad_norm": 1.4825433492660522,
+      "learning_rate": 1.8794871794871797e-05,
+      "loss": 0.9434,
+      "step": 1675
+    },
+    {
+      "epoch": 1.8722466960352424,
+      "grad_norm": 2.4470510482788086,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.8763,
+      "step": 1700
+    },
+    {
+      "epoch": 1.8997797356828194,
+      "grad_norm": 1.226882815361023,
+      "learning_rate": 1.8743589743589744e-05,
+      "loss": 0.8782,
+      "step": 1725
+    },
+    {
+      "epoch": 1.9273127753303965,
+      "grad_norm": 1.1586799621582031,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 0.8885,
+      "step": 1750
+    },
+    {
+      "epoch": 1.9548458149779737,
+      "grad_norm": 2.1358814239501953,
+      "learning_rate": 1.8692307692307695e-05,
+      "loss": 0.8911,
+      "step": 1775
+    },
+    {
+      "epoch": 1.9823788546255505,
+      "grad_norm": 1.2554844617843628,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.8718,
+      "step": 1800
+    },
+    {
+      "epoch": 2.0099118942731278,
+      "grad_norm": 1.0800217390060425,
+      "learning_rate": 1.8641025641025642e-05,
+      "loss": 0.8163,
+      "step": 1825
+    },
+    {
+      "epoch": 2.037444933920705,
+      "grad_norm": 1.413068413734436,
+      "learning_rate": 1.8615384615384616e-05,
+      "loss": 0.6879,
+      "step": 1850
+    },
+    {
+      "epoch": 2.064977973568282,
+      "grad_norm": 1.199027419090271,
+      "learning_rate": 1.8589743589743593e-05,
+      "loss": 0.6669,
+      "step": 1875
+    },
+    {
+      "epoch": 2.092511013215859,
+      "grad_norm": 0.9802116751670837,
+      "learning_rate": 1.8564102564102567e-05,
+      "loss": 0.6631,
+      "step": 1900
+    },
+    {
+      "epoch": 2.1200440528634363,
+      "grad_norm": 1.1249160766601562,
+      "learning_rate": 1.853846153846154e-05,
+      "loss": 0.7245,
+      "step": 1925
+    },
+    {
+      "epoch": 2.147577092511013,
+      "grad_norm": 1.1133592128753662,
+      "learning_rate": 1.8512820512820514e-05,
+      "loss": 0.646,
+      "step": 1950
+    },
+    {
+      "epoch": 2.1751101321585904,
+      "grad_norm": 0.9653272032737732,
+      "learning_rate": 1.848717948717949e-05,
+      "loss": 0.6674,
+      "step": 1975
+    },
+    {
+      "epoch": 2.202643171806167,
+      "grad_norm": 0.9605304002761841,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.6729,
+      "step": 2000
+    },
+    {
+      "epoch": 2.202643171806167,
+      "eval_cer": 36.00758887118407,
+      "eval_loss": 0.9211130738258362,
+      "eval_runtime": 725.5723,
+      "eval_samples_per_second": 14.583,
+      "eval_steps_per_second": 3.647,
+      "eval_wer": 94.93635077793495,
+      "step": 2000
+    },
+    {
+      "epoch": 2.2301762114537445,
+      "grad_norm": 1.5967915058135986,
+      "learning_rate": 1.8435897435897435e-05,
+      "loss": 0.6661,
+      "step": 2025
+    },
+    {
+      "epoch": 2.2577092511013217,
+      "grad_norm": 1.0568780899047852,
+      "learning_rate": 1.8410256410256412e-05,
+      "loss": 0.6776,
+      "step": 2050
+    },
+    {
+      "epoch": 2.2852422907488985,
+      "grad_norm": 0.981234073638916,
+      "learning_rate": 1.8384615384615386e-05,
+      "loss": 0.6297,
+      "step": 2075
+    },
+    {
+      "epoch": 2.3127753303964758,
+      "grad_norm": 1.799085259437561,
+      "learning_rate": 1.835897435897436e-05,
+      "loss": 0.6518,
+      "step": 2100
+    },
+    {
+      "epoch": 2.340308370044053,
+      "grad_norm": 1.3996877670288086,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.6158,
+      "step": 2125
+    },
+    {
+      "epoch": 2.36784140969163,
+      "grad_norm": 1.4223437309265137,
+      "learning_rate": 1.830769230769231e-05,
+      "loss": 0.6413,
+      "step": 2150
+    },
+    {
+      "epoch": 2.395374449339207,
+      "grad_norm": 0.9599340558052063,
+      "learning_rate": 1.8282051282051284e-05,
+      "loss": 0.6391,
+      "step": 2175
+    },
+    {
+      "epoch": 2.4229074889867843,
+      "grad_norm": 1.0826489925384521,
+      "learning_rate": 1.8256410256410257e-05,
+      "loss": 0.6207,
+      "step": 2200
+    },
+    {
+      "epoch": 2.450440528634361,
+      "grad_norm": 1.255297303199768,
+      "learning_rate": 1.823076923076923e-05,
+      "loss": 0.6033,
+      "step": 2225
+    },
+    {
+      "epoch": 2.4779735682819384,
+      "grad_norm": 1.2208808660507202,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 0.6331,
+      "step": 2250
+    },
+    {
+      "epoch": 2.505506607929515,
+      "grad_norm": 0.898612916469574,
+      "learning_rate": 1.817948717948718e-05,
+      "loss": 0.6794,
+      "step": 2275
+    },
+    {
+      "epoch": 2.5330396475770924,
+      "grad_norm": 1.1741799116134644,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.6322,
+      "step": 2300
+    },
+    {
+      "epoch": 2.5605726872246697,
+      "grad_norm": 1.1415616273880005,
+      "learning_rate": 1.812820512820513e-05,
+      "loss": 0.6263,
+      "step": 2325
+    },
+    {
+      "epoch": 2.5881057268722465,
+      "grad_norm": 0.988058865070343,
+      "learning_rate": 1.8102564102564102e-05,
+      "loss": 0.6351,
+      "step": 2350
+    },
+    {
+      "epoch": 2.6156387665198237,
+      "grad_norm": 0.9876915812492371,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.629,
+      "step": 2375
+    },
+    {
+      "epoch": 2.643171806167401,
+      "grad_norm": 1.1506072282791138,
+      "learning_rate": 1.8051282051282053e-05,
+      "loss": 0.598,
+      "step": 2400
+    },
+    {
+      "epoch": 2.670704845814978,
+      "grad_norm": 0.9459457397460938,
+      "learning_rate": 1.8025641025641027e-05,
+      "loss": 0.5881,
+      "step": 2425
+    },
+    {
+      "epoch": 2.698237885462555,
+      "grad_norm": 1.0385560989379883,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6527,
+      "step": 2450
+    },
+    {
+      "epoch": 2.7257709251101323,
+      "grad_norm": 1.0275869369506836,
+      "learning_rate": 1.7974358974358977e-05,
+      "loss": 0.5693,
+      "step": 2475
+    },
+    {
+      "epoch": 2.753303964757709,
+      "grad_norm": 1.068461537361145,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.6406,
+      "step": 2500
+    },
+    {
+      "epoch": 2.7808370044052864,
+      "grad_norm": 0.8605853915214539,
+      "learning_rate": 1.7923076923076925e-05,
+      "loss": 0.6138,
+      "step": 2525
+    },
+    {
+      "epoch": 2.8083700440528636,
+      "grad_norm": 1.1672123670578003,
+      "learning_rate": 1.78974358974359e-05,
+      "loss": 0.5752,
+      "step": 2550
+    },
+    {
+      "epoch": 2.8359030837004404,
+      "grad_norm": 1.1020793914794922,
+      "learning_rate": 1.7871794871794875e-05,
+      "loss": 0.6057,
+      "step": 2575
+    },
+    {
+      "epoch": 2.8634361233480177,
+      "grad_norm": 0.9963453412055969,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.6324,
+      "step": 2600
+    },
+    {
+      "epoch": 2.890969162995595,
+      "grad_norm": 0.9259366989135742,
+      "learning_rate": 1.7820512820512823e-05,
+      "loss": 0.5967,
+      "step": 2625
+    },
+    {
+      "epoch": 2.9185022026431717,
+      "grad_norm": 0.9898194074630737,
+      "learning_rate": 1.7794871794871796e-05,
+      "loss": 0.5975,
+      "step": 2650
+    },
+    {
+      "epoch": 2.946035242290749,
+      "grad_norm": 0.9436148405075073,
+      "learning_rate": 1.776923076923077e-05,
+      "loss": 0.6018,
+      "step": 2675
+    },
+    {
+      "epoch": 2.9735682819383262,
+      "grad_norm": 1.2582800388336182,
+      "learning_rate": 1.7743589743589744e-05,
+      "loss": 0.58,
+      "step": 2700
+    },
+    {
+      "epoch": 3.001101321585903,
+      "grad_norm": 0.7944229245185852,
+      "learning_rate": 1.7717948717948717e-05,
+      "loss": 0.6239,
+      "step": 2725
+    },
+    {
+      "epoch": 3.0286343612334803,
+      "grad_norm": 0.9075860977172852,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.4649,
+      "step": 2750
+    },
+    {
+      "epoch": 3.056167400881057,
+      "grad_norm": 1.5187524557113647,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 0.4216,
+      "step": 2775
+    },
+    {
+      "epoch": 3.0837004405286343,
+      "grad_norm": 0.9071484804153442,
+      "learning_rate": 1.7641025641025642e-05,
+      "loss": 0.4377,
+      "step": 2800
+    },
+    {
+      "epoch": 3.1112334801762116,
+      "grad_norm": 0.5699480772018433,
+      "learning_rate": 1.7615384615384615e-05,
+      "loss": 0.4108,
+      "step": 2825
+    },
+    {
+      "epoch": 3.1387665198237884,
+      "grad_norm": 0.8742319941520691,
+      "learning_rate": 1.7589743589743592e-05,
+      "loss": 0.4213,
+      "step": 2850
+    },
+    {
+      "epoch": 3.1662995594713657,
+      "grad_norm": 0.9574070572853088,
+      "learning_rate": 1.7564102564102566e-05,
+      "loss": 0.4529,
+      "step": 2875
+    },
+    {
+      "epoch": 3.193832599118943,
+      "grad_norm": 0.9607532620429993,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.4519,
+      "step": 2900
+    },
+    {
+      "epoch": 3.2213656387665197,
+      "grad_norm": 0.8327065706253052,
+      "learning_rate": 1.7512820512820513e-05,
+      "loss": 0.4539,
+      "step": 2925
+    },
+    {
+      "epoch": 3.248898678414097,
+      "grad_norm": 0.8266808390617371,
+      "learning_rate": 1.7487179487179487e-05,
+      "loss": 0.4381,
+      "step": 2950
+    },
+    {
+      "epoch": 3.2764317180616738,
+      "grad_norm": 1.3495973348617554,
+      "learning_rate": 1.7461538461538464e-05,
+      "loss": 0.4255,
+      "step": 2975
+    },
+    {
+      "epoch": 3.303964757709251,
+      "grad_norm": 0.781352698802948,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.4087,
+      "step": 3000
+    },
+    {
+      "epoch": 3.303964757709251,
+      "eval_cer": 33.05916339000149,
+      "eval_loss": 0.8169652223587036,
+      "eval_runtime": 726.2485,
+      "eval_samples_per_second": 14.569,
+      "eval_steps_per_second": 3.643,
+      "eval_wer": 91.47571900047147,
+      "step": 3000
+    },
+    {
+      "epoch": 3.3314977973568283,
+      "grad_norm": 0.8976072669029236,
+      "learning_rate": 1.741025641025641e-05,
+      "loss": 0.4375,
+      "step": 3025
+    },
+    {
+      "epoch": 3.359030837004405,
+      "grad_norm": 0.9293546080589294,
+      "learning_rate": 1.7384615384615385e-05,
+      "loss": 0.4645,
+      "step": 3050
+    },
+    {
+      "epoch": 3.3865638766519823,
+      "grad_norm": 0.8952093124389648,
+      "learning_rate": 1.7358974358974362e-05,
+      "loss": 0.4217,
+      "step": 3075
+    },
+    {
+      "epoch": 3.4140969162995596,
+      "grad_norm": 0.9495489597320557,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4842,
+      "step": 3100
+    },
+    {
+      "epoch": 3.4416299559471364,
+      "grad_norm": 0.9268874526023865,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.4474,
+      "step": 3125
+    },
+    {
+      "epoch": 3.4691629955947136,
+      "grad_norm": 1.2101101875305176,
+      "learning_rate": 1.7282051282051283e-05,
+      "loss": 0.4241,
+      "step": 3150
+    },
+    {
+      "epoch": 3.496696035242291,
+      "grad_norm": 0.9907839894294739,
+      "learning_rate": 1.725641025641026e-05,
+      "loss": 0.445,
+      "step": 3175
+    },
+    {
+      "epoch": 3.5242290748898677,
+      "grad_norm": 0.9446345567703247,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.4501,
+      "step": 3200
+    },
+    {
+      "epoch": 3.551762114537445,
+      "grad_norm": 0.8528251051902771,
+      "learning_rate": 1.7205128205128207e-05,
+      "loss": 0.4304,
+      "step": 3225
+    },
+    {
+      "epoch": 3.579295154185022,
+      "grad_norm": 0.8596051335334778,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 0.4218,
+      "step": 3250
+    },
+    {
+      "epoch": 3.606828193832599,
+      "grad_norm": 0.7655468583106995,
+      "learning_rate": 1.7153846153846155e-05,
+      "loss": 0.4416,
+      "step": 3275
+    },
+    {
+      "epoch": 3.6343612334801763,
+      "grad_norm": 0.9918355941772461,
+      "learning_rate": 1.7128205128205128e-05,
+      "loss": 0.452,
+      "step": 3300
+    },
+    {
+      "epoch": 3.6618942731277535,
+      "grad_norm": 0.6974489688873291,
+      "learning_rate": 1.7102564102564102e-05,
+      "loss": 0.3897,
+      "step": 3325
+    },
+    {
+      "epoch": 3.6894273127753303,
+      "grad_norm": 1.001642107963562,
+      "learning_rate": 1.707692307692308e-05,
+      "loss": 0.4241,
+      "step": 3350
+    },
+    {
+      "epoch": 3.7169603524229076,
+      "grad_norm": 1.0877162218093872,
+      "learning_rate": 1.7051282051282053e-05,
+      "loss": 0.3882,
+      "step": 3375
+    },
+    {
+      "epoch": 3.744493392070485,
+      "grad_norm": 0.8201428055763245,
+      "learning_rate": 1.7025641025641026e-05,
+      "loss": 0.4318,
+      "step": 3400
+    },
+    {
+      "epoch": 3.7720264317180616,
+      "grad_norm": 0.9629787802696228,
+      "learning_rate": 1.7e-05,
+      "loss": 0.4128,
+      "step": 3425
+    },
+    {
+      "epoch": 3.799559471365639,
+      "grad_norm": 0.8146516680717468,
+      "learning_rate": 1.6974358974358977e-05,
+      "loss": 0.4122,
+      "step": 3450
+    },
+    {
+      "epoch": 3.827092511013216,
+      "grad_norm": 0.9245569109916687,
+      "learning_rate": 1.694871794871795e-05,
+      "loss": 0.4031,
+      "step": 3475
+    },
+    {
+      "epoch": 3.854625550660793,
+      "grad_norm": 0.8833574652671814,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.479,
+      "step": 3500
+    },
+    {
+      "epoch": 3.88215859030837,
+      "grad_norm": 0.7967398166656494,
+      "learning_rate": 1.6897435897435898e-05,
+      "loss": 0.4076,
+      "step": 3525
+    },
+    {
+      "epoch": 3.909691629955947,
+      "grad_norm": 0.8753622770309448,
+      "learning_rate": 1.687179487179487e-05,
+      "loss": 0.4233,
+      "step": 3550
+    },
+    {
+      "epoch": 3.9372246696035242,
+      "grad_norm": 0.8417605757713318,
+      "learning_rate": 1.684615384615385e-05,
+      "loss": 0.4512,
+      "step": 3575
+    },
+    {
+      "epoch": 3.964757709251101,
+      "grad_norm": 0.8642501831054688,
+      "learning_rate": 1.6820512820512822e-05,
+      "loss": 0.4442,
+      "step": 3600
+    },
+    {
+      "epoch": 3.9922907488986783,
+      "grad_norm": 1.236092209815979,
+      "learning_rate": 1.6794871794871796e-05,
+      "loss": 0.4411,
+      "step": 3625
+    },
+    {
+      "epoch": 4.0198237885462555,
+      "grad_norm": 1.4108121395111084,
+      "learning_rate": 1.676923076923077e-05,
+      "loss": 0.3219,
+      "step": 3650
+    },
+    {
+      "epoch": 4.047356828193832,
+      "grad_norm": 1.2637825012207031,
+      "learning_rate": 1.6743589743589747e-05,
+      "loss": 0.3046,
+      "step": 3675
+    },
+    {
+      "epoch": 4.07488986784141,
+      "grad_norm": 0.7874261736869812,
+      "learning_rate": 1.671794871794872e-05,
+      "loss": 0.2738,
+      "step": 3700
+    },
+    {
+      "epoch": 4.102422907488987,
+      "grad_norm": 0.6468262672424316,
+      "learning_rate": 1.6692307692307694e-05,
+      "loss": 0.3136,
+      "step": 3725
+    },
+    {
+      "epoch": 4.129955947136564,
+      "grad_norm": 0.8033445477485657,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.3052,
+      "step": 3750
+    },
+    {
+      "epoch": 4.157488986784141,
+      "grad_norm": 0.8105751872062683,
+      "learning_rate": 1.6641025641025645e-05,
+      "loss": 0.2836,
+      "step": 3775
+    },
+    {
+      "epoch": 4.185022026431718,
+      "grad_norm": 0.6188549995422363,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.2522,
+      "step": 3800
+    },
+    {
+      "epoch": 4.212555066079295,
+      "grad_norm": 0.697938859462738,
+      "learning_rate": 1.6589743589743592e-05,
+      "loss": 0.2917,
+      "step": 3825
+    },
+    {
+      "epoch": 4.240088105726873,
+      "grad_norm": 0.7022973895072937,
+      "learning_rate": 1.6564102564102565e-05,
+      "loss": 0.2803,
+      "step": 3850
+    },
+    {
+      "epoch": 4.2676211453744495,
+      "grad_norm": 0.7884867191314697,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.3195,
+      "step": 3875
+    },
+    {
+      "epoch": 4.295154185022026,
+      "grad_norm": 0.6753418445587158,
+      "learning_rate": 1.6512820512820513e-05,
+      "loss": 0.309,
+      "step": 3900
+    },
+    {
+      "epoch": 4.322687224669604,
+      "grad_norm": 0.822180986404419,
+      "learning_rate": 1.6487179487179486e-05,
+      "loss": 0.3211,
+      "step": 3925
+    },
+    {
+      "epoch": 4.350220264317181,
+      "grad_norm": 0.8807072639465332,
+      "learning_rate": 1.6461538461538463e-05,
+      "loss": 0.2958,
+      "step": 3950
+    },
+    {
+      "epoch": 4.377753303964758,
+      "grad_norm": 0.6565355062484741,
+      "learning_rate": 1.6435897435897437e-05,
+      "loss": 0.3209,
+      "step": 3975
+    },
+    {
+      "epoch": 4.405286343612334,
+      "grad_norm": 0.910705029964447,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.303,
+      "step": 4000
+    },
+    {
+      "epoch": 4.405286343612334,
+      "eval_cer": 35.491155869911616,
+      "eval_loss": 0.7996189594268799,
+      "eval_runtime": 729.2489,
+      "eval_samples_per_second": 14.509,
+      "eval_steps_per_second": 3.628,
+      "eval_wer": 95.96416784535596,
+      "step": 4000
+    },
+    {
+      "epoch": 4.432819383259912,
+      "grad_norm": 1.0489437580108643,
+      "learning_rate": 1.6384615384615384e-05,
+      "loss": 0.3326,
+      "step": 4025
+    },
+    {
+      "epoch": 4.460352422907489,
+      "grad_norm": 0.9662296175956726,
+      "learning_rate": 1.635897435897436e-05,
+      "loss": 0.332,
+      "step": 4050
+    },
+    {
+      "epoch": 4.487885462555066,
+      "grad_norm": 0.8950140476226807,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 0.3279,
+      "step": 4075
+    },
+    {
+      "epoch": 4.515418502202643,
+      "grad_norm": 0.7907012104988098,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.2761,
+      "step": 4100
+    },
+    {
+      "epoch": 4.54295154185022,
+      "grad_norm": 0.723153829574585,
+      "learning_rate": 1.6282051282051282e-05,
+      "loss": 0.3136,
+      "step": 4125
+    },
+    {
+      "epoch": 4.570484581497797,
+      "grad_norm": 0.7711037397384644,
+      "learning_rate": 1.625641025641026e-05,
+      "loss": 0.3254,
+      "step": 4150
+    },
+    {
+      "epoch": 4.598017621145375,
+      "grad_norm": 0.8112226128578186,
+      "learning_rate": 1.6230769230769233e-05,
+      "loss": 0.3089,
+      "step": 4175
+    },
+    {
+      "epoch": 4.6255506607929515,
+      "grad_norm": 0.6841309070587158,
+      "learning_rate": 1.6205128205128207e-05,
+      "loss": 0.2948,
+      "step": 4200
+    },
+    {
+      "epoch": 4.653083700440528,
+      "grad_norm": 0.8256305456161499,
+      "learning_rate": 1.617948717948718e-05,
+      "loss": 0.3045,
+      "step": 4225
+    },
+    {
+      "epoch": 4.680616740088106,
+      "grad_norm": 0.8179898262023926,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.2873,
+      "step": 4250
+    },
+    {
+      "epoch": 4.708149779735683,
+      "grad_norm": 0.6983807682991028,
+      "learning_rate": 1.612820512820513e-05,
+      "loss": 0.2981,
+      "step": 4275
+    },
+    {
+      "epoch": 4.73568281938326,
+      "grad_norm": 0.9936075806617737,
+      "learning_rate": 1.6102564102564105e-05,
+      "loss": 0.31,
+      "step": 4300
+    },
+    {
+      "epoch": 4.763215859030837,
+      "grad_norm": 0.7161526679992676,
+      "learning_rate": 1.607692307692308e-05,
+      "loss": 0.303,
+      "step": 4325
+    },
+    {
+      "epoch": 4.790748898678414,
+      "grad_norm": 0.7693957686424255,
+      "learning_rate": 1.6051282051282052e-05,
+      "loss": 0.3067,
+      "step": 4350
+    },
+    {
+      "epoch": 4.818281938325991,
+      "grad_norm": 0.8263546824455261,
+      "learning_rate": 1.602564102564103e-05,
+      "loss": 0.3104,
+      "step": 4375
+    },
+    {
+      "epoch": 4.845814977973569,
+      "grad_norm": 0.540008544921875,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2855,
+      "step": 4400
+    },
+    {
+      "epoch": 4.8733480176211454,
+      "grad_norm": 0.7057309746742249,
+      "learning_rate": 1.5974358974358976e-05,
+      "loss": 0.2847,
+      "step": 4425
+    },
+    {
+      "epoch": 4.900881057268722,
+      "grad_norm": 0.7432449460029602,
+      "learning_rate": 1.594871794871795e-05,
+      "loss": 0.303,
+      "step": 4450
+    },
+    {
+      "epoch": 4.9284140969163,
+      "grad_norm": 0.7021868824958801,
+      "learning_rate": 1.5923076923076924e-05,
+      "loss": 0.2833,
+      "step": 4475
+    },
+    {
+      "epoch": 4.955947136563877,
+      "grad_norm": 0.7491558194160461,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.2936,
+      "step": 4500
+    },
+    {
+      "epoch": 4.983480176211454,
+      "grad_norm": 0.6752347350120544,
+      "learning_rate": 1.587179487179487e-05,
+      "loss": 0.3019,
+      "step": 4525
+    },
+    {
+      "epoch": 5.011013215859031,
+      "grad_norm": 0.6153631806373596,
+      "learning_rate": 1.5846153846153848e-05,
+      "loss": 0.2446,
+      "step": 4550
+    },
+    {
+      "epoch": 5.038546255506608,
+      "grad_norm": 0.3978467881679535,
+      "learning_rate": 1.582051282051282e-05,
+      "loss": 0.2114,
+      "step": 4575
+    },
+    {
+      "epoch": 5.066079295154185,
+      "grad_norm": 0.6924332976341248,
+      "learning_rate": 1.5794871794871795e-05,
+      "loss": 0.1749,
+      "step": 4600
+    },
+    {
+      "epoch": 5.093612334801762,
+      "grad_norm": 0.5500088930130005,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.2242,
+      "step": 4625
+    },
+    {
+      "epoch": 5.121145374449339,
+      "grad_norm": 0.515316367149353,
+      "learning_rate": 1.5743589743589746e-05,
+      "loss": 0.2148,
+      "step": 4650
+    },
+    {
+      "epoch": 5.148678414096916,
+      "grad_norm": 0.446718692779541,
+      "learning_rate": 1.571794871794872e-05,
+      "loss": 0.2136,
+      "step": 4675
+    },
+    {
+      "epoch": 5.176211453744493,
+      "grad_norm": 0.48106926679611206,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.1853,
+      "step": 4700
+    },
+    {
+      "epoch": 5.203744493392071,
+      "grad_norm": 0.49837005138397217,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 0.2107,
+      "step": 4725
+    },
+    {
+      "epoch": 5.2312775330396475,
+      "grad_norm": 0.9187563061714172,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 0.1851,
+      "step": 4750
+    },
+    {
+      "epoch": 5.258810572687224,
+      "grad_norm": 0.6566996574401855,
+      "learning_rate": 1.5615384615384618e-05,
+      "loss": 0.2419,
+      "step": 4775
+    },
+    {
+      "epoch": 5.286343612334802,
+      "grad_norm": 0.48451924324035645,
+      "learning_rate": 1.558974358974359e-05,
+      "loss": 0.1932,
+      "step": 4800
+    },
+    {
+      "epoch": 5.313876651982379,
+      "grad_norm": 0.43492308259010315,
+      "learning_rate": 1.5564102564102565e-05,
+      "loss": 0.1785,
+      "step": 4825
+    },
+    {
+      "epoch": 5.341409691629956,
+      "grad_norm": 0.5435590147972107,
+      "learning_rate": 1.553846153846154e-05,
+      "loss": 0.1986,
+      "step": 4850
+    },
+    {
+      "epoch": 5.368942731277533,
+      "grad_norm": 0.5883561372756958,
+      "learning_rate": 1.5512820512820516e-05,
+      "loss": 0.2044,
+      "step": 4875
+    },
+    {
+      "epoch": 5.39647577092511,
+      "grad_norm": 0.7071317434310913,
+      "learning_rate": 1.548717948717949e-05,
+      "loss": 0.1901,
+      "step": 4900
+    },
+    {
+      "epoch": 5.424008810572687,
+      "grad_norm": 0.6903939247131348,
+      "learning_rate": 1.5461538461538463e-05,
+      "loss": 0.2427,
+      "step": 4925
+    },
+    {
+      "epoch": 5.451541850220265,
+      "grad_norm": 0.8299788236618042,
+      "learning_rate": 1.5435897435897436e-05,
+      "loss": 0.2219,
+      "step": 4950
+    },
+    {
+      "epoch": 5.479074889867841,
+      "grad_norm": 0.7349267601966858,
+      "learning_rate": 1.5410256410256414e-05,
+      "loss": 0.2021,
+      "step": 4975
+    },
+    {
+      "epoch": 5.506607929515418,
+      "grad_norm": 0.7333081960678101,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.211,
+      "step": 5000
+    },
+    {
+      "epoch": 5.506607929515418,
+      "eval_cer": 39.8708057730446,
+      "eval_loss": 0.7910177707672119,
+      "eval_runtime": 793.4489,
+      "eval_samples_per_second": 13.335,
+      "eval_steps_per_second": 3.335,
+      "eval_wer": 90.76850542197077,
+      "step": 5000
+    },
+    {
+      "epoch": 5.534140969162996,
+      "grad_norm": 0.6456535458564758,
+      "learning_rate": 1.535897435897436e-05,
+      "loss": 0.2052,
+      "step": 5025
+    },
+    {
+      "epoch": 5.561674008810573,
+      "grad_norm": 0.5858263969421387,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.2287,
+      "step": 5050
+    },
+    {
+      "epoch": 5.5892070484581495,
+      "grad_norm": 0.8086832165718079,
+      "learning_rate": 1.5307692307692308e-05,
+      "loss": 0.2084,
+      "step": 5075
+    },
+    {
+      "epoch": 5.616740088105727,
+      "grad_norm": 0.7293650507926941,
+      "learning_rate": 1.5282051282051282e-05,
+      "loss": 0.2037,
+      "step": 5100
+    },
+    {
+      "epoch": 5.644273127753304,
+      "grad_norm": 0.7970709800720215,
+      "learning_rate": 1.5256410256410257e-05,
+      "loss": 0.2093,
+      "step": 5125
+    },
+    {
+      "epoch": 5.671806167400881,
+      "grad_norm": 0.6010832786560059,
+      "learning_rate": 1.523076923076923e-05,
+      "loss": 0.224,
+      "step": 5150
+    },
+    {
+      "epoch": 5.6993392070484585,
+      "grad_norm": 0.6445685625076294,
+      "learning_rate": 1.5205128205128206e-05,
+      "loss": 0.184,
+      "step": 5175
+    },
+    {
+      "epoch": 5.726872246696035,
+      "grad_norm": 0.6301259994506836,
+      "learning_rate": 1.517948717948718e-05,
+      "loss": 0.2357,
+      "step": 5200
+    },
+    {
+      "epoch": 5.754405286343612,
+      "grad_norm": 0.6070804595947266,
+      "learning_rate": 1.5153846153846155e-05,
+      "loss": 0.2029,
+      "step": 5225
+    },
+    {
+      "epoch": 5.78193832599119,
+      "grad_norm": 0.7481094002723694,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.2419,
+      "step": 5250
+    },
+    {
+      "epoch": 5.809471365638767,
+      "grad_norm": 0.6122421622276306,
+      "learning_rate": 1.5102564102564104e-05,
+      "loss": 0.2123,
+      "step": 5275
+    },
+    {
+      "epoch": 5.8370044052863435,
+      "grad_norm": 0.7841750383377075,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.2072,
+      "step": 5300
+    },
+    {
+      "epoch": 5.864537444933921,
+      "grad_norm": 0.7243798971176147,
+      "learning_rate": 1.5051282051282053e-05,
+      "loss": 0.2082,
+      "step": 5325
+    },
+    {
+      "epoch": 5.892070484581498,
+      "grad_norm": 0.7279136180877686,
+      "learning_rate": 1.5025641025641027e-05,
+      "loss": 0.2209,
+      "step": 5350
+    },
+    {
+      "epoch": 5.919603524229075,
+      "grad_norm": 0.5851007699966431,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1917,
+      "step": 5375
+    },
+    {
+      "epoch": 5.9471365638766525,
+      "grad_norm": 0.7515987157821655,
+      "learning_rate": 1.4974358974358976e-05,
+      "loss": 0.201,
+      "step": 5400
+    },
+    {
+      "epoch": 5.974669603524229,
+      "grad_norm": 0.7539699077606201,
+      "learning_rate": 1.494871794871795e-05,
+      "loss": 0.1881,
+      "step": 5425
+    },
+    {
+      "epoch": 6.002202643171806,
+      "grad_norm": 0.5623589158058167,
+      "learning_rate": 1.4923076923076925e-05,
+      "loss": 0.1731,
+      "step": 5450
+    },
+    {
+      "epoch": 6.029735682819383,
+      "grad_norm": 0.5110183358192444,
+      "learning_rate": 1.4897435897435898e-05,
+      "loss": 0.1283,
+      "step": 5475
+    },
+    {
+      "epoch": 6.057268722466961,
+      "grad_norm": 0.48959916830062866,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.1116,
+      "step": 5500
+    },
+    {
+      "epoch": 6.084801762114537,
+      "grad_norm": 0.5873674750328064,
+      "learning_rate": 1.4846153846153847e-05,
+      "loss": 0.1396,
+      "step": 5525
+    },
+    {
+      "epoch": 6.112334801762114,
+      "grad_norm": 0.6229241490364075,
+      "learning_rate": 1.4820512820512823e-05,
+      "loss": 0.1478,
+      "step": 5550
+    },
+    {
+      "epoch": 6.139867841409692,
+      "grad_norm": 0.39609479904174805,
+      "learning_rate": 1.4794871794871796e-05,
+      "loss": 0.1335,
+      "step": 5575
+    },
+    {
+      "epoch": 6.167400881057269,
+      "grad_norm": 0.5592531561851501,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.1574,
+      "step": 5600
+    },
+    {
+      "epoch": 6.1949339207048455,
+      "grad_norm": 0.4108933210372925,
+      "learning_rate": 1.4743589743589745e-05,
+      "loss": 0.1082,
+      "step": 5625
+    },
+    {
+      "epoch": 6.222466960352423,
+      "grad_norm": 0.7125318646430969,
+      "learning_rate": 1.471794871794872e-05,
+      "loss": 0.1455,
+      "step": 5650
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.6028311252593994,
+      "learning_rate": 1.4692307692307694e-05,
+      "loss": 0.1476,
+      "step": 5675
+    },
+    {
+      "epoch": 6.277533039647577,
+      "grad_norm": 0.3542225956916809,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.1148,
+      "step": 5700
+    },
+    {
+      "epoch": 6.3050660792951545,
+      "grad_norm": 0.3824445307254791,
+      "learning_rate": 1.4641025641025642e-05,
+      "loss": 0.1366,
+      "step": 5725
+    },
+    {
+      "epoch": 6.332599118942731,
+      "grad_norm": 0.4482337534427643,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.1176,
+      "step": 5750
+    },
+    {
+      "epoch": 6.360132158590308,
+      "grad_norm": 0.6524671316146851,
+      "learning_rate": 1.458974358974359e-05,
+      "loss": 0.1275,
+      "step": 5775
+    },
+    {
+      "epoch": 6.387665198237886,
+      "grad_norm": 0.6443710327148438,
+      "learning_rate": 1.4564102564102564e-05,
+      "loss": 0.1425,
+      "step": 5800
+    },
+    {
+      "epoch": 6.415198237885463,
+      "grad_norm": 0.4919815957546234,
+      "learning_rate": 1.453846153846154e-05,
+      "loss": 0.1277,
+      "step": 5825
+    },
+    {
+      "epoch": 6.442731277533039,
+      "grad_norm": 0.6433417201042175,
+      "learning_rate": 1.4512820512820513e-05,
+      "loss": 0.1519,
+      "step": 5850
+    },
+    {
+      "epoch": 6.470264317180617,
+      "grad_norm": 0.687128484249115,
+      "learning_rate": 1.4487179487179489e-05,
+      "loss": 0.1572,
+      "step": 5875
+    },
+    {
+      "epoch": 6.497797356828194,
+      "grad_norm": 0.5720986127853394,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.1427,
+      "step": 5900
+    },
+    {
+      "epoch": 6.525330396475771,
+      "grad_norm": 0.5719828009605408,
+      "learning_rate": 1.4435897435897438e-05,
+      "loss": 0.155,
+      "step": 5925
+    },
+    {
+      "epoch": 6.5528634361233475,
+      "grad_norm": 1.1837940216064453,
+      "learning_rate": 1.4410256410256411e-05,
+      "loss": 0.1468,
+      "step": 5950
+    },
+    {
+      "epoch": 6.580396475770925,
+      "grad_norm": 0.5758116245269775,
+      "learning_rate": 1.4384615384615387e-05,
+      "loss": 0.1384,
+      "step": 5975
+    },
+    {
+      "epoch": 6.607929515418502,
+      "grad_norm": 0.46409040689468384,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.1389,
+      "step": 6000
+    },
+    {
+      "epoch": 6.607929515418502,
+      "eval_cer": 46.33109029839625,
+      "eval_loss": 0.813330888748169,
+      "eval_runtime": 879.8873,
+      "eval_samples_per_second": 12.025,
+      "eval_steps_per_second": 3.007,
+      "eval_wer": 91.25884016973126,
+      "step": 6000
+    },
+    {
+      "epoch": 6.635462555066079,
+      "grad_norm": 0.6982928514480591,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 0.1128,
+      "step": 6025
+    },
+    {
+      "epoch": 6.6629955947136565,
+      "grad_norm": 0.4645637273788452,
+      "learning_rate": 1.430769230769231e-05,
+      "loss": 0.1611,
+      "step": 6050
+    },
+    {
+      "epoch": 6.690528634361233,
+      "grad_norm": 0.7097769379615784,
+      "learning_rate": 1.4282051282051283e-05,
+      "loss": 0.1686,
+      "step": 6075
+    },
+    {
+      "epoch": 6.71806167400881,
+      "grad_norm": 0.6409720182418823,
+      "learning_rate": 1.4256410256410258e-05,
+      "loss": 0.1645,
+      "step": 6100
+    },
+    {
+      "epoch": 6.745594713656388,
+      "grad_norm": 0.4065110683441162,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.1556,
+      "step": 6125
+    },
+    {
+      "epoch": 6.773127753303965,
+      "grad_norm": 0.5424155592918396,
+      "learning_rate": 1.4205128205128207e-05,
+      "loss": 0.1433,
+      "step": 6150
+    },
+    {
+      "epoch": 6.8006607929515415,
+      "grad_norm": 0.547636091709137,
+      "learning_rate": 1.4179487179487181e-05,
+      "loss": 0.1419,
+      "step": 6175
+    },
+    {
+      "epoch": 6.828193832599119,
+      "grad_norm": 0.6900294423103333,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.1317,
+      "step": 6200
+    },
+    {
+      "epoch": 6.855726872246696,
+      "grad_norm": 0.5622290372848511,
+      "learning_rate": 1.412820512820513e-05,
+      "loss": 0.1604,
+      "step": 6225
+    },
+    {
+      "epoch": 6.883259911894273,
+      "grad_norm": 0.47399094700813293,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 0.1352,
+      "step": 6250
+    },
+    {
+      "epoch": 6.9107929515418505,
+      "grad_norm": 0.5894147157669067,
+      "learning_rate": 1.4076923076923079e-05,
+      "loss": 0.1316,
+      "step": 6275
+    },
+    {
+      "epoch": 6.938325991189427,
+      "grad_norm": 0.5824093222618103,
+      "learning_rate": 1.405128205128205e-05,
+      "loss": 0.1425,
+      "step": 6300
+    },
+    {
+      "epoch": 6.965859030837004,
+      "grad_norm": 0.6649166345596313,
+      "learning_rate": 1.4025641025641026e-05,
+      "loss": 0.147,
+      "step": 6325
+    },
+    {
+      "epoch": 6.993392070484582,
+      "grad_norm": 0.6389392018318176,
+      "learning_rate": 1.4e-05,
+      "loss": 0.1444,
+      "step": 6350
+    },
+    {
+      "epoch": 7.020925110132159,
+      "grad_norm": 0.2867998480796814,
+      "learning_rate": 1.3974358974358975e-05,
+      "loss": 0.0929,
+      "step": 6375
+    },
+    {
+      "epoch": 7.048458149779735,
+      "grad_norm": 0.4151160717010498,
+      "learning_rate": 1.3948717948717949e-05,
+      "loss": 0.0895,
+      "step": 6400
+    },
+    {
+      "epoch": 7.075991189427313,
+      "grad_norm": 0.28654003143310547,
+      "learning_rate": 1.3923076923076924e-05,
+      "loss": 0.0773,
+      "step": 6425
+    },
+    {
+      "epoch": 7.10352422907489,
+      "grad_norm": 0.40104398131370544,
+      "learning_rate": 1.3897435897435898e-05,
+      "loss": 0.0717,
+      "step": 6450
+    },
+    {
+      "epoch": 7.131057268722467,
+      "grad_norm": 0.3279833495616913,
+      "learning_rate": 1.3871794871794873e-05,
+      "loss": 0.0896,
+      "step": 6475
+    },
+    {
+      "epoch": 7.158590308370044,
+      "grad_norm": 0.5332910418510437,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.099,
+      "step": 6500
+    },
+    {
+      "epoch": 7.186123348017621,
+      "grad_norm": 0.5952982902526855,
+      "learning_rate": 1.3820512820512822e-05,
+      "loss": 0.0917,
+      "step": 6525
+    },
+    {
+      "epoch": 7.213656387665198,
+      "grad_norm": 0.6255844831466675,
+      "learning_rate": 1.3794871794871796e-05,
+      "loss": 0.0912,
+      "step": 6550
+    },
+    {
+      "epoch": 7.241189427312776,
+      "grad_norm": 0.26723840832710266,
+      "learning_rate": 1.3769230769230771e-05,
+      "loss": 0.0916,
+      "step": 6575
+    },
+    {
+      "epoch": 7.2687224669603525,
+      "grad_norm": 0.6658725738525391,
+      "learning_rate": 1.3743589743589745e-05,
+      "loss": 0.0843,
+      "step": 6600
+    },
+    {
+      "epoch": 7.296255506607929,
+      "grad_norm": 0.7884182333946228,
+      "learning_rate": 1.3717948717948718e-05,
+      "loss": 0.1023,
+      "step": 6625
+    },
+    {
+      "epoch": 7.323788546255507,
+      "grad_norm": 0.3756254017353058,
+      "learning_rate": 1.3692307692307694e-05,
+      "loss": 0.0919,
+      "step": 6650
+    },
+    {
+      "epoch": 7.351321585903084,
+      "grad_norm": 0.4542330205440521,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 0.1014,
+      "step": 6675
+    },
+    {
+      "epoch": 7.378854625550661,
+      "grad_norm": 0.4459224343299866,
+      "learning_rate": 1.3641025641025643e-05,
+      "loss": 0.0799,
+      "step": 6700
+    },
+    {
+      "epoch": 7.406387665198238,
+      "grad_norm": 0.27258846163749695,
+      "learning_rate": 1.3615384615384616e-05,
+      "loss": 0.0781,
+      "step": 6725
+    },
+    {
+      "epoch": 7.433920704845815,
+      "grad_norm": 0.3643723130226135,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 0.0927,
+      "step": 6750
+    },
+    {
+      "epoch": 7.461453744493392,
+      "grad_norm": 0.4624654948711395,
+      "learning_rate": 1.3564102564102565e-05,
+      "loss": 0.104,
+      "step": 6775
+    },
+    {
+      "epoch": 7.48898678414097,
+      "grad_norm": 0.7944092750549316,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.0994,
+      "step": 6800
+    },
+    {
+      "epoch": 7.516519823788546,
+      "grad_norm": 0.47494593262672424,
+      "learning_rate": 1.3512820512820514e-05,
+      "loss": 0.0905,
+      "step": 6825
+    },
+    {
+      "epoch": 7.544052863436123,
+      "grad_norm": 0.38293811678886414,
+      "learning_rate": 1.348717948717949e-05,
+      "loss": 0.0881,
+      "step": 6850
+    },
+    {
+      "epoch": 7.5715859030837,
+      "grad_norm": 0.43859726190567017,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.0844,
+      "step": 6875
+    },
+    {
+      "epoch": 7.599118942731278,
+      "grad_norm": 0.4808247983455658,
+      "learning_rate": 1.3435897435897435e-05,
+      "loss": 0.1076,
+      "step": 6900
+    },
+    {
+      "epoch": 7.6266519823788546,
+      "grad_norm": 0.2447817623615265,
+      "learning_rate": 1.341025641025641e-05,
+      "loss": 0.0839,
+      "step": 6925
+    },
+    {
+      "epoch": 7.654185022026431,
+      "grad_norm": 0.587374210357666,
+      "learning_rate": 1.3384615384615384e-05,
+      "loss": 0.1139,
+      "step": 6950
+    },
+    {
+      "epoch": 7.681718061674009,
+      "grad_norm": 0.5492860078811646,
+      "learning_rate": 1.335897435897436e-05,
+      "loss": 0.1034,
+      "step": 6975
+    },
+    {
+      "epoch": 7.709251101321586,
+      "grad_norm": 0.5353982448577881,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0864,
+      "step": 7000
+    },
+    {
+      "epoch": 7.709251101321586,
+      "eval_cer": 39.91780633475864,
+      "eval_loss": 0.8311607837677002,
+      "eval_runtime": 794.0532,
+      "eval_samples_per_second": 13.325,
+      "eval_steps_per_second": 3.332,
+      "eval_wer": 92.66383781235267,
+      "step": 7000
+    },
+    {
+      "epoch": 7.736784140969163,
+      "grad_norm": 0.6064012050628662,
+      "learning_rate": 1.3307692307692309e-05,
+      "loss": 0.1134,
+      "step": 7025
+    },
+    {
+      "epoch": 7.76431718061674,
+      "grad_norm": 0.4052427411079407,
+      "learning_rate": 1.3282051282051282e-05,
+      "loss": 0.0944,
+      "step": 7050
+    },
+    {
+      "epoch": 7.791850220264317,
+      "grad_norm": 0.5817214846611023,
+      "learning_rate": 1.3256410256410258e-05,
+      "loss": 0.0905,
+      "step": 7075
+    },
+    {
+      "epoch": 7.819383259911894,
+      "grad_norm": 0.566748321056366,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.0856,
+      "step": 7100
+    },
+    {
+      "epoch": 7.846916299559472,
+      "grad_norm": 0.568682074546814,
+      "learning_rate": 1.3205128205128207e-05,
+      "loss": 0.0793,
+      "step": 7125
+    },
+    {
+      "epoch": 7.8744493392070485,
+      "grad_norm": 0.3249245882034302,
+      "learning_rate": 1.317948717948718e-05,
+      "loss": 0.1105,
+      "step": 7150
+    },
+    {
+      "epoch": 7.901982378854625,
+      "grad_norm": 0.33808597922325134,
+      "learning_rate": 1.3153846153846156e-05,
+      "loss": 0.1038,
+      "step": 7175
+    },
+    {
+      "epoch": 7.929515418502203,
+      "grad_norm": 0.7098983526229858,
+      "learning_rate": 1.312820512820513e-05,
+      "loss": 0.1197,
+      "step": 7200
+    },
+    {
+      "epoch": 7.95704845814978,
+      "grad_norm": 0.5019989609718323,
+      "learning_rate": 1.3102564102564103e-05,
+      "loss": 0.1096,
+      "step": 7225
+    },
+    {
+      "epoch": 7.984581497797357,
+      "grad_norm": 0.3687189817428589,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.104,
+      "step": 7250
+    },
+    {
+      "epoch": 8.012114537444933,
+      "grad_norm": 0.3454643785953522,
+      "learning_rate": 1.3051282051282052e-05,
+      "loss": 0.1124,
+      "step": 7275
+    },
+    {
+      "epoch": 8.039647577092511,
+      "grad_norm": 0.4261622428894043,
+      "learning_rate": 1.3025641025641027e-05,
+      "loss": 0.0676,
+      "step": 7300
+    },
+    {
+      "epoch": 8.067180616740089,
+      "grad_norm": 0.6200215220451355,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.0654,
+      "step": 7325
+    },
+    {
+      "epoch": 8.094713656387665,
+      "grad_norm": 0.2601291239261627,
+      "learning_rate": 1.2974358974358976e-05,
+      "loss": 0.0684,
+      "step": 7350
+    },
+    {
+      "epoch": 8.122246696035242,
+      "grad_norm": 0.43284061551094055,
+      "learning_rate": 1.294871794871795e-05,
+      "loss": 0.0666,
+      "step": 7375
+    },
+    {
+      "epoch": 8.14977973568282,
+      "grad_norm": 0.3861032724380493,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.062,
+      "step": 7400
+    },
+    {
+      "epoch": 8.177312775330396,
+      "grad_norm": 0.20073507726192474,
+      "learning_rate": 1.2897435897435899e-05,
+      "loss": 0.0634,
+      "step": 7425
+    },
+    {
+      "epoch": 8.204845814977974,
+      "grad_norm": 0.16386602818965912,
+      "learning_rate": 1.2871794871794874e-05,
+      "loss": 0.0558,
+      "step": 7450
+    },
+    {
+      "epoch": 8.232378854625551,
+      "grad_norm": 0.5257241129875183,
+      "learning_rate": 1.2846153846153848e-05,
+      "loss": 0.0798,
+      "step": 7475
+    },
+    {
+      "epoch": 8.259911894273127,
+      "grad_norm": 0.2548966407775879,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.056,
+      "step": 7500
+    },
+    {
+      "epoch": 8.287444933920705,
+      "grad_norm": 0.5919742584228516,
+      "learning_rate": 1.2794871794871795e-05,
+      "loss": 0.0608,
+      "step": 7525
+    },
+    {
+      "epoch": 8.314977973568283,
+      "grad_norm": 0.6152638792991638,
+      "learning_rate": 1.2769230769230769e-05,
+      "loss": 0.0592,
+      "step": 7550
+    },
+    {
+      "epoch": 8.342511013215859,
+      "grad_norm": 0.6119657158851624,
+      "learning_rate": 1.2743589743589744e-05,
+      "loss": 0.071,
+      "step": 7575
+    },
+    {
+      "epoch": 8.370044052863436,
+      "grad_norm": 0.2810268998146057,
+      "learning_rate": 1.2717948717948718e-05,
+      "loss": 0.0561,
+      "step": 7600
+    },
+    {
+      "epoch": 8.397577092511014,
+      "grad_norm": 0.5555744171142578,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.0744,
+      "step": 7625
+    },
+    {
+      "epoch": 8.42511013215859,
+      "grad_norm": 0.3237517476081848,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.0563,
+      "step": 7650
+    },
+    {
+      "epoch": 8.452643171806168,
+      "grad_norm": 0.48900526762008667,
+      "learning_rate": 1.2641025641025642e-05,
+      "loss": 0.0635,
+      "step": 7675
+    },
+    {
+      "epoch": 8.480176211453745,
+      "grad_norm": 0.22445560991764069,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.0565,
+      "step": 7700
+    },
+    {
+      "epoch": 8.507709251101321,
+      "grad_norm": 0.2636544108390808,
+      "learning_rate": 1.2589743589743591e-05,
+      "loss": 0.0562,
+      "step": 7725
+    },
+    {
+      "epoch": 8.535242290748899,
+      "grad_norm": 0.5531124472618103,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.0739,
+      "step": 7750
+    },
+    {
+      "epoch": 8.562775330396477,
+      "grad_norm": 0.3963180482387543,
+      "learning_rate": 1.253846153846154e-05,
+      "loss": 0.0723,
+      "step": 7775
+    },
+    {
+      "epoch": 8.590308370044053,
+      "grad_norm": 0.39044126868247986,
+      "learning_rate": 1.2512820512820514e-05,
+      "loss": 0.0709,
+      "step": 7800
+    },
+    {
+      "epoch": 8.61784140969163,
+      "grad_norm": 0.17790181934833527,
+      "learning_rate": 1.2487179487179487e-05,
+      "loss": 0.0761,
+      "step": 7825
+    },
+    {
+      "epoch": 8.645374449339208,
+      "grad_norm": 0.4327470064163208,
+      "learning_rate": 1.2461538461538463e-05,
+      "loss": 0.0558,
+      "step": 7850
+    },
+    {
+      "epoch": 8.672907488986784,
+      "grad_norm": 0.40538233518600464,
+      "learning_rate": 1.2435897435897436e-05,
+      "loss": 0.0525,
+      "step": 7875
+    },
+    {
+      "epoch": 8.700440528634362,
+      "grad_norm": 0.6931684613227844,
+      "learning_rate": 1.2410256410256412e-05,
+      "loss": 0.0714,
+      "step": 7900
+    },
+    {
+      "epoch": 8.72797356828194,
+      "grad_norm": 0.34026893973350525,
+      "learning_rate": 1.2384615384615385e-05,
+      "loss": 0.0672,
+      "step": 7925
+    },
+    {
+      "epoch": 8.755506607929515,
+      "grad_norm": 0.5128918886184692,
+      "learning_rate": 1.235897435897436e-05,
+      "loss": 0.0683,
+      "step": 7950
+    },
+    {
+      "epoch": 8.783039647577093,
+      "grad_norm": 0.6940311789512634,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 0.0662,
+      "step": 7975
+    },
+    {
+      "epoch": 8.810572687224669,
+      "grad_norm": 0.6918107271194458,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.0729,
+      "step": 8000
+    },
+    {
+      "epoch": 8.810572687224669,
+      "eval_cer": 50.74341132369631,
+      "eval_loss": 0.8530433773994446,
+      "eval_runtime": 917.533,
+      "eval_samples_per_second": 11.532,
+      "eval_steps_per_second": 2.884,
+      "eval_wer": 91.61716171617161,
+      "step": 8000
+    },
+    {
+      "epoch": 8.838105726872246,
+      "grad_norm": 0.41443535685539246,
+      "learning_rate": 1.2282051282051283e-05,
+      "loss": 0.0601,
+      "step": 8025
+    },
+    {
+      "epoch": 8.865638766519824,
+      "grad_norm": 0.7258812785148621,
+      "learning_rate": 1.2256410256410259e-05,
+      "loss": 0.0713,
+      "step": 8050
+    },
+    {
+      "epoch": 8.8931718061674,
+      "grad_norm": 0.4442296326160431,
+      "learning_rate": 1.2230769230769232e-05,
+      "loss": 0.0656,
+      "step": 8075
+    },
+    {
+      "epoch": 8.920704845814978,
+      "grad_norm": 0.2696678638458252,
+      "learning_rate": 1.2205128205128208e-05,
+      "loss": 0.0781,
+      "step": 8100
+    },
+    {
+      "epoch": 8.948237885462555,
+      "grad_norm": 0.37608885765075684,
+      "learning_rate": 1.217948717948718e-05,
+      "loss": 0.0617,
+      "step": 8125
+    },
+    {
+      "epoch": 8.975770925110131,
+      "grad_norm": 0.34984931349754333,
+      "learning_rate": 1.2153846153846153e-05,
+      "loss": 0.052,
+      "step": 8150
+    },
+    {
+      "epoch": 9.003303964757709,
+      "grad_norm": 0.19837628304958344,
+      "learning_rate": 1.2128205128205129e-05,
+      "loss": 0.0659,
+      "step": 8175
+    },
+    {
+      "epoch": 9.030837004405287,
+      "grad_norm": 0.3520928621292114,
+      "learning_rate": 1.2102564102564102e-05,
+      "loss": 0.0356,
+      "step": 8200
+    },
+    {
+      "epoch": 9.058370044052863,
+      "grad_norm": 0.1488259732723236,
+      "learning_rate": 1.2076923076923078e-05,
+      "loss": 0.0365,
+      "step": 8225
+    },
+    {
+      "epoch": 9.08590308370044,
+      "grad_norm": 0.5500857830047607,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.0328,
+      "step": 8250
+    },
+    {
+      "epoch": 9.113436123348018,
+      "grad_norm": 0.4940767288208008,
+      "learning_rate": 1.2025641025641027e-05,
+      "loss": 0.0385,
+      "step": 8275
+    },
+    {
+      "epoch": 9.140969162995594,
+      "grad_norm": 0.499952495098114,
+      "learning_rate": 1.2e-05,
+      "loss": 0.0441,
+      "step": 8300
+    },
+    {
+      "epoch": 9.168502202643172,
+      "grad_norm": 0.28660815954208374,
+      "learning_rate": 1.1974358974358976e-05,
+      "loss": 0.0302,
+      "step": 8325
+    },
+    {
+      "epoch": 9.19603524229075,
+      "grad_norm": 0.29444143176078796,
+      "learning_rate": 1.194871794871795e-05,
+      "loss": 0.0435,
+      "step": 8350
+    },
+    {
+      "epoch": 9.223568281938325,
+      "grad_norm": 0.17032678425312042,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.0383,
+      "step": 8375
+    },
+    {
+      "epoch": 9.251101321585903,
+      "grad_norm": 0.2526102364063263,
+      "learning_rate": 1.1897435897435898e-05,
+      "loss": 0.0392,
+      "step": 8400
+    },
+    {
+      "epoch": 9.27863436123348,
+      "grad_norm": 0.21672683954238892,
+      "learning_rate": 1.1871794871794872e-05,
+      "loss": 0.0368,
+      "step": 8425
+    },
+    {
+      "epoch": 9.306167400881057,
+      "grad_norm": 0.17891252040863037,
+      "learning_rate": 1.1846153846153847e-05,
+      "loss": 0.045,
+      "step": 8450
+    },
+    {
+      "epoch": 9.333700440528634,
+      "grad_norm": 0.29890188574790955,
+      "learning_rate": 1.1820512820512821e-05,
+      "loss": 0.0374,
+      "step": 8475
+    },
+    {
+      "epoch": 9.361233480176212,
+      "grad_norm": 0.2425035685300827,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.0463,
+      "step": 8500
+    },
+    {
+      "epoch": 9.388766519823788,
+      "grad_norm": 0.1933910995721817,
+      "learning_rate": 1.176923076923077e-05,
+      "loss": 0.0423,
+      "step": 8525
+    },
+    {
+      "epoch": 9.416299559471366,
+      "grad_norm": 0.26760435104370117,
+      "learning_rate": 1.1743589743589745e-05,
+      "loss": 0.0483,
+      "step": 8550
+    },
+    {
+      "epoch": 9.443832599118943,
+      "grad_norm": 0.2656070291996002,
+      "learning_rate": 1.1717948717948719e-05,
+      "loss": 0.0413,
+      "step": 8575
+    },
+    {
+      "epoch": 9.47136563876652,
+      "grad_norm": 0.4633423686027527,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.0402,
+      "step": 8600
+    },
+    {
+      "epoch": 9.498898678414097,
+      "grad_norm": 0.52508544921875,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.052,
+      "step": 8625
+    },
+    {
+      "epoch": 9.526431718061675,
+      "grad_norm": 0.5961471796035767,
+      "learning_rate": 1.1641025641025643e-05,
+      "loss": 0.0533,
+      "step": 8650
+    },
+    {
+      "epoch": 9.55396475770925,
+      "grad_norm": 0.25339117646217346,
+      "learning_rate": 1.1615384615384617e-05,
+      "loss": 0.0457,
+      "step": 8675
+    },
+    {
+      "epoch": 9.581497797356828,
+      "grad_norm": 0.3615815043449402,
+      "learning_rate": 1.1589743589743592e-05,
+      "loss": 0.0363,
+      "step": 8700
+    },
+    {
+      "epoch": 9.609030837004406,
+      "grad_norm": 0.17995741963386536,
+      "learning_rate": 1.1564102564102566e-05,
+      "loss": 0.0501,
+      "step": 8725
+    },
+    {
+      "epoch": 9.636563876651982,
+      "grad_norm": 0.5352070927619934,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.0513,
+      "step": 8750
+    },
+    {
+      "epoch": 9.66409691629956,
+      "grad_norm": 0.48025304079055786,
+      "learning_rate": 1.1512820512820513e-05,
+      "loss": 0.043,
+      "step": 8775
+    },
+    {
+      "epoch": 9.691629955947137,
+      "grad_norm": 0.45890381932258606,
+      "learning_rate": 1.1487179487179487e-05,
+      "loss": 0.0389,
+      "step": 8800
+    },
+    {
+      "epoch": 9.719162995594713,
+      "grad_norm": 0.572181761264801,
+      "learning_rate": 1.1461538461538462e-05,
+      "loss": 0.0487,
+      "step": 8825
+    },
+    {
+      "epoch": 9.746696035242291,
+      "grad_norm": 0.382355272769928,
+      "learning_rate": 1.1435897435897436e-05,
+      "loss": 0.0527,
+      "step": 8850
+    },
+    {
+      "epoch": 9.774229074889869,
+      "grad_norm": 0.3971492052078247,
+      "learning_rate": 1.1410256410256411e-05,
+      "loss": 0.0493,
+      "step": 8875
+    },
+    {
+      "epoch": 9.801762114537445,
+      "grad_norm": 0.520282506942749,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.0389,
+      "step": 8900
+    },
+    {
+      "epoch": 9.829295154185022,
+      "grad_norm": 0.2088589370250702,
+      "learning_rate": 1.135897435897436e-05,
+      "loss": 0.052,
+      "step": 8925
+    },
+    {
+      "epoch": 9.8568281938326,
+      "grad_norm": 0.569845974445343,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.043,
+      "step": 8950
+    },
+    {
+      "epoch": 9.884361233480176,
+      "grad_norm": 0.4208398759365082,
+      "learning_rate": 1.1307692307692309e-05,
+      "loss": 0.0378,
+      "step": 8975
+    },
+    {
+      "epoch": 9.911894273127754,
+      "grad_norm": 0.24408133327960968,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.0381,
+      "step": 9000
+    },
+    {
+      "epoch": 9.911894273127754,
+      "eval_cer": 47.71588733621451,
+      "eval_loss": 0.8698176741600037,
+      "eval_runtime": 865.0989,
+      "eval_samples_per_second": 12.231,
+      "eval_steps_per_second": 3.059,
+      "eval_wer": 91.57001414427157,
+      "step": 9000
+    },
+    {
+      "epoch": 9.939427312775331,
+      "grad_norm": 0.8426101207733154,
+      "learning_rate": 1.1256410256410258e-05,
+      "loss": 0.0433,
+      "step": 9025
+    },
+    {
+      "epoch": 9.966960352422907,
+      "grad_norm": 0.5663429498672485,
+      "learning_rate": 1.1230769230769232e-05,
+      "loss": 0.0461,
+      "step": 9050
+    },
+    {
+      "epoch": 9.994493392070485,
+      "grad_norm": 0.4001486003398895,
+      "learning_rate": 1.1205128205128205e-05,
+      "loss": 0.0508,
+      "step": 9075
+    },
+    {
+      "epoch": 10.022026431718063,
+      "grad_norm": 0.11667744815349579,
+      "learning_rate": 1.117948717948718e-05,
+      "loss": 0.0281,
+      "step": 9100
+    },
+    {
+      "epoch": 10.049559471365638,
+      "grad_norm": 0.3421935439109802,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.0277,
+      "step": 9125
+    },
+    {
+      "epoch": 10.077092511013216,
+      "grad_norm": 0.39479511976242065,
+      "learning_rate": 1.112820512820513e-05,
+      "loss": 0.0341,
+      "step": 9150
+    },
+    {
+      "epoch": 10.104625550660794,
+      "grad_norm": 0.5420315861701965,
+      "learning_rate": 1.1102564102564103e-05,
+      "loss": 0.0249,
+      "step": 9175
+    },
+    {
+      "epoch": 10.13215859030837,
+      "grad_norm": 0.4009842276573181,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.0277,
+      "step": 9200
+    },
+    {
+      "epoch": 10.159691629955947,
+      "grad_norm": 0.14988014101982117,
+      "learning_rate": 1.1051282051282052e-05,
+      "loss": 0.0254,
+      "step": 9225
+    },
+    {
+      "epoch": 10.187224669603523,
+      "grad_norm": 0.10612168908119202,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.0293,
+      "step": 9250
+    },
+    {
+      "epoch": 10.214757709251101,
+      "grad_norm": 0.07644204050302505,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.02,
+      "step": 9275
+    },
+    {
+      "epoch": 10.242290748898679,
+      "grad_norm": 0.37732774019241333,
+      "learning_rate": 1.0974358974358977e-05,
+      "loss": 0.0252,
+      "step": 9300
+    },
+    {
+      "epoch": 10.269823788546255,
+      "grad_norm": 0.362834095954895,
+      "learning_rate": 1.094871794871795e-05,
+      "loss": 0.0287,
+      "step": 9325
+    },
+    {
+      "epoch": 10.297356828193832,
+      "grad_norm": 0.37372374534606934,
+      "learning_rate": 1.0923076923076922e-05,
+      "loss": 0.0279,
+      "step": 9350
+    },
+    {
+      "epoch": 10.32488986784141,
+      "grad_norm": 0.1827729046344757,
+      "learning_rate": 1.0897435897435898e-05,
+      "loss": 0.0383,
+      "step": 9375
+    },
+    {
+      "epoch": 10.352422907488986,
+      "grad_norm": 0.79392009973526,
+      "learning_rate": 1.0871794871794871e-05,
+      "loss": 0.0218,
+      "step": 9400
+    },
+    {
+      "epoch": 10.379955947136564,
+      "grad_norm": 0.4169488251209259,
+      "learning_rate": 1.0846153846153847e-05,
+      "loss": 0.0331,
+      "step": 9425
+    },
+    {
+      "epoch": 10.407488986784141,
+      "grad_norm": 0.197611466050148,
+      "learning_rate": 1.082051282051282e-05,
+      "loss": 0.0285,
+      "step": 9450
+    },
+    {
+      "epoch": 10.435022026431717,
+      "grad_norm": 0.20039740204811096,
+      "learning_rate": 1.0794871794871796e-05,
+      "loss": 0.0219,
+      "step": 9475
+    },
+    {
+      "epoch": 10.462555066079295,
+      "grad_norm": 0.5181693434715271,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.031,
+      "step": 9500
+    },
+    {
+      "epoch": 10.490088105726873,
+      "grad_norm": 0.31759241223335266,
+      "learning_rate": 1.0743589743589745e-05,
+      "loss": 0.0386,
+      "step": 9525
+    },
+    {
+      "epoch": 10.517621145374449,
+      "grad_norm": 0.20832429826259613,
+      "learning_rate": 1.0717948717948718e-05,
+      "loss": 0.0266,
+      "step": 9550
+    },
+    {
+      "epoch": 10.545154185022026,
+      "grad_norm": 0.1244155541062355,
+      "learning_rate": 1.0692307692307694e-05,
+      "loss": 0.0282,
+      "step": 9575
+    },
+    {
+      "epoch": 10.572687224669604,
+      "grad_norm": 0.1275760680437088,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.026,
+      "step": 9600
+    },
+    {
+      "epoch": 10.60022026431718,
+      "grad_norm": 0.5201808214187622,
+      "learning_rate": 1.0641025641025643e-05,
+      "loss": 0.0284,
+      "step": 9625
+    },
+    {
+      "epoch": 10.627753303964758,
+      "grad_norm": 0.6102043390274048,
+      "learning_rate": 1.0615384615384616e-05,
+      "loss": 0.0279,
+      "step": 9650
+    },
+    {
+      "epoch": 10.655286343612335,
+      "grad_norm": 0.19192653894424438,
+      "learning_rate": 1.058974358974359e-05,
+      "loss": 0.0317,
+      "step": 9675
+    },
+    {
+      "epoch": 10.682819383259911,
+      "grad_norm": 0.2622509300708771,
+      "learning_rate": 1.0564102564102565e-05,
+      "loss": 0.0356,
+      "step": 9700
+    },
+    {
+      "epoch": 10.710352422907489,
+      "grad_norm": 0.16083978116512299,
+      "learning_rate": 1.0538461538461539e-05,
+      "loss": 0.0318,
+      "step": 9725
+    },
+    {
+      "epoch": 10.737885462555067,
+      "grad_norm": 0.29113614559173584,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.0301,
+      "step": 9750
+    },
+    {
+      "epoch": 10.765418502202643,
+      "grad_norm": 0.1394374966621399,
+      "learning_rate": 1.0487179487179488e-05,
+      "loss": 0.0249,
+      "step": 9775
+    },
+    {
+      "epoch": 10.79295154185022,
+      "grad_norm": 0.45710182189941406,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.0326,
+      "step": 9800
+    },
+    {
+      "epoch": 10.820484581497798,
+      "grad_norm": 0.3842594027519226,
+      "learning_rate": 1.0435897435897437e-05,
+      "loss": 0.028,
+      "step": 9825
+    },
+    {
+      "epoch": 10.848017621145374,
+      "grad_norm": 0.1398581862449646,
+      "learning_rate": 1.0410256410256412e-05,
+      "loss": 0.0258,
+      "step": 9850
+    },
+    {
+      "epoch": 10.875550660792952,
+      "grad_norm": 0.30973750352859497,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.0287,
+      "step": 9875
+    },
+    {
+      "epoch": 10.90308370044053,
+      "grad_norm": 0.4438304901123047,
+      "learning_rate": 1.0358974358974361e-05,
+      "loss": 0.0208,
+      "step": 9900
+    },
+    {
+      "epoch": 10.930616740088105,
+      "grad_norm": 0.3679158389568329,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 0.0193,
+      "step": 9925
+    },
+    {
+      "epoch": 10.958149779735683,
+      "grad_norm": 0.19606485962867737,
+      "learning_rate": 1.0307692307692307e-05,
+      "loss": 0.0281,
+      "step": 9950
+    },
+    {
+      "epoch": 10.98568281938326,
+      "grad_norm": 0.2789362967014313,
+      "learning_rate": 1.0282051282051282e-05,
+      "loss": 0.033,
+      "step": 9975
+    },
+    {
+      "epoch": 11.013215859030836,
+      "grad_norm": 0.4207296669483185,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.028,
+      "step": 10000
+    },
+    {
+      "epoch": 11.013215859030836,
+      "eval_cer": 54.37563766005984,
+      "eval_loss": 0.886362612247467,
+      "eval_runtime": 981.5076,
+      "eval_samples_per_second": 10.78,
+      "eval_steps_per_second": 2.696,
+      "eval_wer": 92.14521452145215,
+      "step": 10000
+    },
+    {
+      "epoch": 11.040748898678414,
+      "grad_norm": 0.2317088097333908,
+      "learning_rate": 1.0230769230769231e-05,
+      "loss": 0.0188,
+      "step": 10025
+    },
+    {
+      "epoch": 11.068281938325992,
+      "grad_norm": 0.1417902708053589,
+      "learning_rate": 1.0205128205128205e-05,
+      "loss": 0.0215,
+      "step": 10050
+    },
+    {
+      "epoch": 11.095814977973568,
+      "grad_norm": 0.07895007729530334,
+      "learning_rate": 1.017948717948718e-05,
+      "loss": 0.0206,
+      "step": 10075
+    },
+    {
+      "epoch": 11.123348017621145,
+      "grad_norm": 0.23651820421218872,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.0193,
+      "step": 10100
+    },
+    {
+      "epoch": 11.150881057268723,
+      "grad_norm": 0.30718091130256653,
+      "learning_rate": 1.012820512820513e-05,
+      "loss": 0.0181,
+      "step": 10125
+    },
+    {
+      "epoch": 11.178414096916299,
+      "grad_norm": 0.5389750003814697,
+      "learning_rate": 1.0102564102564103e-05,
+      "loss": 0.0227,
+      "step": 10150
+    },
+    {
+      "epoch": 11.205947136563877,
+      "grad_norm": 0.1611407846212387,
+      "learning_rate": 1.0076923076923078e-05,
+      "loss": 0.0179,
+      "step": 10175
+    },
+    {
+      "epoch": 11.233480176211454,
+      "grad_norm": 0.12234646826982498,
+      "learning_rate": 1.0051282051282052e-05,
+      "loss": 0.0177,
+      "step": 10200
+    },
+    {
+      "epoch": 11.26101321585903,
+      "grad_norm": 0.17735643684864044,
+      "learning_rate": 1.0025641025641027e-05,
+      "loss": 0.018,
+      "step": 10225
+    },
+    {
+      "epoch": 11.288546255506608,
+      "grad_norm": 0.4353015720844269,
+      "learning_rate": 1e-05,
+      "loss": 0.0148,
+      "step": 10250
+    },
+    {
+      "epoch": 11.316079295154186,
+      "grad_norm": 0.5004972219467163,
+      "learning_rate": 9.974358974358974e-06,
+      "loss": 0.0226,
+      "step": 10275
+    },
+    {
+      "epoch": 11.343612334801762,
+      "grad_norm": 0.21323665976524353,
+      "learning_rate": 9.94871794871795e-06,
+      "loss": 0.0175,
+      "step": 10300
+    },
+    {
+      "epoch": 11.37114537444934,
+      "grad_norm": 0.1589115411043167,
+      "learning_rate": 9.923076923076923e-06,
+      "loss": 0.0171,
+      "step": 10325
+    },
+    {
+      "epoch": 11.398678414096917,
+      "grad_norm": 0.09055918455123901,
+      "learning_rate": 9.897435897435899e-06,
+      "loss": 0.0202,
+      "step": 10350
+    },
+    {
+      "epoch": 11.426211453744493,
+      "grad_norm": 0.15250088274478912,
+      "learning_rate": 9.871794871794872e-06,
+      "loss": 0.0165,
+      "step": 10375
+    },
+    {
+      "epoch": 11.45374449339207,
+      "grad_norm": 0.17833657562732697,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.0226,
+      "step": 10400
+    },
+    {
+      "epoch": 11.481277533039648,
+      "grad_norm": 0.29985514283180237,
+      "learning_rate": 9.820512820512821e-06,
+      "loss": 0.0215,
+      "step": 10425
+    },
+    {
+      "epoch": 11.508810572687224,
+      "grad_norm": 0.1902422308921814,
+      "learning_rate": 9.794871794871795e-06,
+      "loss": 0.0202,
+      "step": 10450
+    },
+    {
+      "epoch": 11.536343612334802,
+      "grad_norm": 0.20311668515205383,
+      "learning_rate": 9.76923076923077e-06,
+      "loss": 0.015,
+      "step": 10475
+    },
+    {
+      "epoch": 11.56387665198238,
+      "grad_norm": 0.29688936471939087,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.0165,
+      "step": 10500
+    },
+    {
+      "epoch": 11.591409691629956,
+      "grad_norm": 0.16763679683208466,
+      "learning_rate": 9.71794871794872e-06,
+      "loss": 0.0226,
+      "step": 10525
+    },
+    {
+      "epoch": 11.618942731277533,
+      "grad_norm": 0.2504405081272125,
+      "learning_rate": 9.692307692307693e-06,
+      "loss": 0.0154,
+      "step": 10550
+    },
+    {
+      "epoch": 11.646475770925111,
+      "grad_norm": 0.4480707347393036,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 0.0197,
+      "step": 10575
+    },
+    {
+      "epoch": 11.674008810572687,
+      "grad_norm": 0.1413351148366928,
+      "learning_rate": 9.641025641025642e-06,
+      "loss": 0.0169,
+      "step": 10600
+    },
+    {
+      "epoch": 11.701541850220265,
+      "grad_norm": 0.32270994782447815,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.0186,
+      "step": 10625
+    },
+    {
+      "epoch": 11.729074889867842,
+      "grad_norm": 0.10015879571437836,
+      "learning_rate": 9.589743589743591e-06,
+      "loss": 0.0193,
+      "step": 10650
+    },
+    {
+      "epoch": 11.756607929515418,
+      "grad_norm": 0.2840658128261566,
+      "learning_rate": 9.564102564102565e-06,
+      "loss": 0.0219,
+      "step": 10675
+    },
+    {
+      "epoch": 11.784140969162996,
+      "grad_norm": 0.3592904508113861,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.0296,
+      "step": 10700
+    },
+    {
+      "epoch": 11.811674008810574,
+      "grad_norm": 0.3074311912059784,
+      "learning_rate": 9.512820512820514e-06,
+      "loss": 0.0219,
+      "step": 10725
+    },
+    {
+      "epoch": 11.83920704845815,
+      "grad_norm": 0.09315291047096252,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.0245,
+      "step": 10750
+    },
+    {
+      "epoch": 11.866740088105727,
+      "grad_norm": 0.23775820434093475,
+      "learning_rate": 9.461538461538463e-06,
+      "loss": 0.0176,
+      "step": 10775
+    },
+    {
+      "epoch": 11.894273127753303,
+      "grad_norm": 0.3934062123298645,
+      "learning_rate": 9.435897435897436e-06,
+      "loss": 0.0193,
+      "step": 10800
+    },
+    {
+      "epoch": 11.92180616740088,
+      "grad_norm": 0.11783640831708908,
+      "learning_rate": 9.410256410256412e-06,
+      "loss": 0.0173,
+      "step": 10825
+    },
+    {
+      "epoch": 11.949339207048459,
+      "grad_norm": 0.2406715750694275,
+      "learning_rate": 9.384615384615385e-06,
+      "loss": 0.0176,
+      "step": 10850
+    },
+    {
+      "epoch": 11.976872246696034,
+      "grad_norm": 0.23296691477298737,
+      "learning_rate": 9.358974358974359e-06,
+      "loss": 0.0225,
+      "step": 10875
+    },
+    {
+      "epoch": 12.004405286343612,
+      "grad_norm": 0.07544398307800293,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.0192,
+      "step": 10900
+    },
+    {
+      "epoch": 12.03193832599119,
+      "grad_norm": 0.11225544661283493,
+      "learning_rate": 9.307692307692308e-06,
+      "loss": 0.0151,
+      "step": 10925
+    },
+    {
+      "epoch": 12.059471365638766,
+      "grad_norm": 0.2554400563240051,
+      "learning_rate": 9.282051282051283e-06,
+      "loss": 0.018,
+      "step": 10950
+    },
+    {
+      "epoch": 12.087004405286343,
+      "grad_norm": 0.278838574886322,
+      "learning_rate": 9.256410256410257e-06,
+      "loss": 0.018,
+      "step": 10975
+    },
+    {
+      "epoch": 12.114537444933921,
+      "grad_norm": 0.15447276830673218,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.0142,
+      "step": 11000
+    },
+    {
+      "epoch": 12.114537444933921,
+      "eval_cer": 53.64139717767359,
+      "eval_loss": 0.8987538814544678,
+      "eval_runtime": 874.0135,
+      "eval_samples_per_second": 12.106,
+      "eval_steps_per_second": 3.027,
+      "eval_wer": 93.21074964639321,
+      "step": 11000
+    },
+    {
+      "epoch": 12.142070484581497,
+      "grad_norm": 0.14940273761749268,
+      "learning_rate": 9.205128205128206e-06,
+      "loss": 0.0134,
+      "step": 11025
+    },
+    {
+      "epoch": 12.169603524229075,
+      "grad_norm": 0.16579824686050415,
+      "learning_rate": 9.17948717948718e-06,
+      "loss": 0.0101,
+      "step": 11050
+    },
+    {
+      "epoch": 12.197136563876652,
+      "grad_norm": 0.19196206331253052,
+      "learning_rate": 9.153846153846155e-06,
+      "loss": 0.0175,
+      "step": 11075
+    },
+    {
+      "epoch": 12.224669603524228,
+      "grad_norm": 0.17125152051448822,
+      "learning_rate": 9.128205128205129e-06,
+      "loss": 0.0146,
+      "step": 11100
+    },
+    {
+      "epoch": 12.252202643171806,
+      "grad_norm": 0.1070375144481659,
+      "learning_rate": 9.102564102564104e-06,
+      "loss": 0.0135,
+      "step": 11125
+    },
+    {
+      "epoch": 12.279735682819384,
+      "grad_norm": 0.18338176608085632,
+      "learning_rate": 9.076923076923078e-06,
+      "loss": 0.0114,
+      "step": 11150
+    },
+    {
+      "epoch": 12.30726872246696,
+      "grad_norm": 0.08506150543689728,
+      "learning_rate": 9.051282051282051e-06,
+      "loss": 0.0109,
+      "step": 11175
+    },
+    {
+      "epoch": 12.334801762114537,
+      "grad_norm": 0.22113458812236786,
+      "learning_rate": 9.025641025641027e-06,
+      "loss": 0.0146,
+      "step": 11200
+    },
+    {
+      "epoch": 12.362334801762115,
+      "grad_norm": 0.7724553942680359,
+      "learning_rate": 9e-06,
+      "loss": 0.0144,
+      "step": 11225
+    },
+    {
+      "epoch": 12.389867841409691,
+      "grad_norm": 0.2879244089126587,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.0158,
+      "step": 11250
+    },
+    {
+      "epoch": 12.417400881057269,
+      "grad_norm": 0.10946013778448105,
+      "learning_rate": 8.94871794871795e-06,
+      "loss": 0.0149,
+      "step": 11275
+    },
+    {
+      "epoch": 12.444933920704846,
+      "grad_norm": 0.16864745318889618,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.0125,
+      "step": 11300
+    },
+    {
+      "epoch": 12.472466960352422,
+      "grad_norm": 0.13665378093719482,
+      "learning_rate": 8.897435897435898e-06,
+      "loss": 0.0125,
+      "step": 11325
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 0.7579739093780518,
+      "learning_rate": 8.871794871794872e-06,
+      "loss": 0.0187,
+      "step": 11350
+    },
+    {
+      "epoch": 12.527533039647578,
+      "grad_norm": 0.2458302229642868,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.0163,
+      "step": 11375
+    },
+    {
+      "epoch": 12.555066079295154,
+      "grad_norm": 0.1165897399187088,
+      "learning_rate": 8.820512820512821e-06,
+      "loss": 0.0203,
+      "step": 11400
+    },
+    {
+      "epoch": 12.582599118942731,
+      "grad_norm": 0.18201911449432373,
+      "learning_rate": 8.794871794871796e-06,
+      "loss": 0.0176,
+      "step": 11425
+    },
+    {
+      "epoch": 12.610132158590309,
+      "grad_norm": 0.24092814326286316,
+      "learning_rate": 8.76923076923077e-06,
+      "loss": 0.0184,
+      "step": 11450
+    },
+    {
+      "epoch": 12.637665198237885,
+      "grad_norm": 0.22921068966388702,
+      "learning_rate": 8.743589743589743e-06,
+      "loss": 0.0165,
+      "step": 11475
+    },
+    {
+      "epoch": 12.665198237885463,
+      "grad_norm": 0.11735259741544724,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.0148,
+      "step": 11500
+    },
+    {
+      "epoch": 12.69273127753304,
+      "grad_norm": 0.3713417053222656,
+      "learning_rate": 8.692307692307692e-06,
+      "loss": 0.013,
+      "step": 11525
+    },
+    {
+      "epoch": 12.720264317180616,
+      "grad_norm": 0.12812820076942444,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.0131,
+      "step": 11550
+    },
+    {
+      "epoch": 12.747797356828194,
+      "grad_norm": 0.08026193827390671,
+      "learning_rate": 8.641025641025641e-06,
+      "loss": 0.0122,
+      "step": 11575
+    },
+    {
+      "epoch": 12.775330396475772,
+      "grad_norm": 0.30850231647491455,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.0162,
+      "step": 11600
+    },
+    {
+      "epoch": 12.802863436123348,
+      "grad_norm": 0.11349613219499588,
+      "learning_rate": 8.58974358974359e-06,
+      "loss": 0.0142,
+      "step": 11625
+    },
+    {
+      "epoch": 12.830396475770925,
+      "grad_norm": 0.0808100774884224,
+      "learning_rate": 8.564102564102564e-06,
+      "loss": 0.0133,
+      "step": 11650
+    },
+    {
+      "epoch": 12.857929515418503,
+      "grad_norm": 0.21294915676116943,
+      "learning_rate": 8.53846153846154e-06,
+      "loss": 0.0131,
+      "step": 11675
+    },
+    {
+      "epoch": 12.885462555066079,
+      "grad_norm": 0.3267536163330078,
+      "learning_rate": 8.512820512820513e-06,
+      "loss": 0.0132,
+      "step": 11700
+    },
+    {
+      "epoch": 12.912995594713657,
+      "grad_norm": 0.1333635449409485,
+      "learning_rate": 8.487179487179488e-06,
+      "loss": 0.0127,
+      "step": 11725
+    },
+    {
+      "epoch": 12.940528634361234,
+      "grad_norm": 0.15748950839042664,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.014,
+      "step": 11750
+    },
+    {
+      "epoch": 12.96806167400881,
+      "grad_norm": 0.10807313024997711,
+      "learning_rate": 8.435897435897436e-06,
+      "loss": 0.0113,
+      "step": 11775
+    },
+    {
+      "epoch": 12.995594713656388,
+      "grad_norm": 0.31078580021858215,
+      "learning_rate": 8.410256410256411e-06,
+      "loss": 0.0143,
+      "step": 11800
+    },
+    {
+      "epoch": 13.023127753303966,
+      "grad_norm": 0.39036718010902405,
+      "learning_rate": 8.384615384615385e-06,
+      "loss": 0.0143,
+      "step": 11825
+    },
+    {
+      "epoch": 13.050660792951541,
+      "grad_norm": 0.2274981439113617,
+      "learning_rate": 8.35897435897436e-06,
+      "loss": 0.0095,
+      "step": 11850
+    },
+    {
+      "epoch": 13.07819383259912,
+      "grad_norm": 0.20098534226417542,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0098,
+      "step": 11875
+    },
+    {
+      "epoch": 13.105726872246697,
+      "grad_norm": 0.06564383208751678,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.0089,
+      "step": 11900
+    },
+    {
+      "epoch": 13.133259911894273,
+      "grad_norm": 0.3132261633872986,
+      "learning_rate": 8.282051282051283e-06,
+      "loss": 0.01,
+      "step": 11925
+    },
+    {
+      "epoch": 13.16079295154185,
+      "grad_norm": 0.2533165514469147,
+      "learning_rate": 8.256410256410256e-06,
+      "loss": 0.0115,
+      "step": 11950
+    },
+    {
+      "epoch": 13.188325991189428,
+      "grad_norm": 0.2507915198802948,
+      "learning_rate": 8.230769230769232e-06,
+      "loss": 0.0119,
+      "step": 11975
+    },
+    {
+      "epoch": 13.215859030837004,
+      "grad_norm": 0.2795217037200928,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.0131,
+      "step": 12000
+    },
+    {
+      "epoch": 13.215859030837004,
+      "eval_cer": 46.21530842685681,
+      "eval_loss": 0.9192056655883789,
+      "eval_runtime": 848.6891,
+      "eval_samples_per_second": 12.467,
+      "eval_steps_per_second": 3.118,
+      "eval_wer": 92.8052805280528,
+      "step": 12000
+    },
+    {
+      "epoch": 13.243392070484582,
+      "grad_norm": 0.08327347785234451,
+      "learning_rate": 8.17948717948718e-06,
+      "loss": 0.0108,
+      "step": 12025
+    },
+    {
+      "epoch": 13.270925110132158,
+      "grad_norm": 0.1631169319152832,
+      "learning_rate": 8.153846153846154e-06,
+      "loss": 0.0073,
+      "step": 12050
+    },
+    {
+      "epoch": 13.298458149779735,
+      "grad_norm": 0.20178398489952087,
+      "learning_rate": 8.12820512820513e-06,
+      "loss": 0.0089,
+      "step": 12075
+    },
+    {
+      "epoch": 13.325991189427313,
+      "grad_norm": 0.1222369447350502,
+      "learning_rate": 8.102564102564103e-06,
+      "loss": 0.0102,
+      "step": 12100
+    },
+    {
+      "epoch": 13.353524229074889,
+      "grad_norm": 0.08848530054092407,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.0132,
+      "step": 12125
+    },
+    {
+      "epoch": 13.381057268722467,
+      "grad_norm": 0.17377710342407227,
+      "learning_rate": 8.051282051282052e-06,
+      "loss": 0.0104,
+      "step": 12150
+    },
+    {
+      "epoch": 13.408590308370044,
+      "grad_norm": 0.2735235095024109,
+      "learning_rate": 8.025641025641026e-06,
+      "loss": 0.009,
+      "step": 12175
+    },
+    {
+      "epoch": 13.43612334801762,
+      "grad_norm": 0.1769276261329651,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.0097,
+      "step": 12200
+    },
+    {
+      "epoch": 13.463656387665198,
+      "grad_norm": 0.11117183417081833,
+      "learning_rate": 7.974358974358975e-06,
+      "loss": 0.0092,
+      "step": 12225
+    },
+    {
+      "epoch": 13.491189427312776,
+      "grad_norm": 0.12082304805517197,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.0111,
+      "step": 12250
+    },
+    {
+      "epoch": 13.518722466960352,
+      "grad_norm": 0.05868387967348099,
+      "learning_rate": 7.923076923076924e-06,
+      "loss": 0.0109,
+      "step": 12275
+    },
+    {
+      "epoch": 13.54625550660793,
+      "grad_norm": 0.6051620244979858,
+      "learning_rate": 7.897435897435898e-06,
+      "loss": 0.0116,
+      "step": 12300
+    },
+    {
+      "epoch": 13.573788546255507,
+      "grad_norm": 0.09658774733543396,
+      "learning_rate": 7.871794871794873e-06,
+      "loss": 0.0106,
+      "step": 12325
+    },
+    {
+      "epoch": 13.601321585903083,
+      "grad_norm": 0.0718512088060379,
+      "learning_rate": 7.846153846153847e-06,
+      "loss": 0.01,
+      "step": 12350
+    },
+    {
+      "epoch": 13.62885462555066,
+      "grad_norm": 0.36982008814811707,
+      "learning_rate": 7.820512820512822e-06,
+      "loss": 0.0113,
+      "step": 12375
+    },
+    {
+      "epoch": 13.656387665198238,
+      "grad_norm": 0.04029988497495651,
+      "learning_rate": 7.794871794871796e-06,
+      "loss": 0.0096,
+      "step": 12400
+    },
+    {
+      "epoch": 13.683920704845814,
+      "grad_norm": 0.05479320138692856,
+      "learning_rate": 7.76923076923077e-06,
+      "loss": 0.0094,
+      "step": 12425
+    },
+    {
+      "epoch": 13.711453744493392,
+      "grad_norm": 0.05368011072278023,
+      "learning_rate": 7.743589743589745e-06,
+      "loss": 0.0088,
+      "step": 12450
+    },
+    {
+      "epoch": 13.73898678414097,
+      "grad_norm": 0.080776646733284,
+      "learning_rate": 7.717948717948718e-06,
+      "loss": 0.0083,
+      "step": 12475
+    },
+    {
+      "epoch": 13.766519823788546,
+      "grad_norm": 0.14407703280448914,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.0095,
+      "step": 12500
+    },
+    {
+      "epoch": 13.794052863436123,
+      "grad_norm": 0.04322797805070877,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 0.0099,
+      "step": 12525
+    },
+    {
+      "epoch": 13.821585903083701,
+      "grad_norm": 0.3888992965221405,
+      "learning_rate": 7.641025641025641e-06,
+      "loss": 0.0101,
+      "step": 12550
+    },
+    {
+      "epoch": 13.849118942731277,
+      "grad_norm": 0.07158606499433517,
+      "learning_rate": 7.615384615384615e-06,
+      "loss": 0.0086,
+      "step": 12575
+    },
+    {
+      "epoch": 13.876651982378855,
+      "grad_norm": 0.044584546238183975,
+      "learning_rate": 7.58974358974359e-06,
+      "loss": 0.0096,
+      "step": 12600
+    },
+    {
+      "epoch": 13.904185022026432,
+      "grad_norm": 0.06555311381816864,
+      "learning_rate": 7.564102564102564e-06,
+      "loss": 0.0106,
+      "step": 12625
+    },
+    {
+      "epoch": 13.931718061674008,
+      "grad_norm": 0.07280756533145905,
+      "learning_rate": 7.538461538461539e-06,
+      "loss": 0.0114,
+      "step": 12650
+    },
+    {
+      "epoch": 13.959251101321586,
+      "grad_norm": 0.09671244770288467,
+      "learning_rate": 7.512820512820513e-06,
+      "loss": 0.0107,
+      "step": 12675
+    },
+    {
+      "epoch": 13.986784140969164,
+      "grad_norm": 0.16803735494613647,
+      "learning_rate": 7.487179487179488e-06,
+      "loss": 0.0132,
+      "step": 12700
+    },
+    {
+      "epoch": 14.01431718061674,
+      "grad_norm": 0.07752186059951782,
+      "learning_rate": 7.461538461538462e-06,
+      "loss": 0.0113,
+      "step": 12725
+    },
+    {
+      "epoch": 14.041850220264317,
+      "grad_norm": 0.1300247758626938,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.0093,
+      "step": 12750
+    },
+    {
+      "epoch": 14.069383259911895,
+      "grad_norm": 0.08789961785078049,
+      "learning_rate": 7.410256410256411e-06,
+      "loss": 0.0073,
+      "step": 12775
+    },
+    {
+      "epoch": 14.09691629955947,
+      "grad_norm": 0.07116591930389404,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.0069,
+      "step": 12800
+    },
+    {
+      "epoch": 14.124449339207048,
+      "grad_norm": 0.09893615543842316,
+      "learning_rate": 7.35897435897436e-06,
+      "loss": 0.0082,
+      "step": 12825
+    },
+    {
+      "epoch": 14.151982378854626,
+      "grad_norm": 0.0769718661904335,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.0088,
+      "step": 12850
+    },
+    {
+      "epoch": 14.179515418502202,
+      "grad_norm": 0.05386563390493393,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.0058,
+      "step": 12875
+    },
+    {
+      "epoch": 14.20704845814978,
+      "grad_norm": 0.11878286302089691,
+      "learning_rate": 7.282051282051282e-06,
+      "loss": 0.0083,
+      "step": 12900
+    },
+    {
+      "epoch": 14.234581497797357,
+      "grad_norm": 0.2804628908634186,
+      "learning_rate": 7.256410256410257e-06,
+      "loss": 0.0099,
+      "step": 12925
+    },
+    {
+      "epoch": 14.262114537444933,
+      "grad_norm": 0.18361924588680267,
+      "learning_rate": 7.230769230769231e-06,
+      "loss": 0.0069,
+      "step": 12950
+    },
+    {
+      "epoch": 14.289647577092511,
+      "grad_norm": 0.23271577060222626,
+      "learning_rate": 7.205128205128206e-06,
+      "loss": 0.0099,
+      "step": 12975
+    },
+    {
+      "epoch": 14.317180616740089,
+      "grad_norm": 0.3786558508872986,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.0088,
+      "step": 13000
+    },
+    {
+      "epoch": 14.317180616740089,
+      "eval_cer": 54.3102954157257,
+      "eval_loss": 0.9230103492736816,
+      "eval_runtime": 846.9631,
+      "eval_samples_per_second": 12.493,
+      "eval_steps_per_second": 3.124,
+      "eval_wer": 93.81423856671381,
+      "step": 13000
+    },
+    {
+      "epoch": 14.344713656387665,
+      "grad_norm": 0.03461921960115433,
+      "learning_rate": 7.153846153846155e-06,
+      "loss": 0.0071,
+      "step": 13025
+    },
+    {
+      "epoch": 14.372246696035242,
+      "grad_norm": 0.06544573605060577,
+      "learning_rate": 7.128205128205129e-06,
+      "loss": 0.0073,
+      "step": 13050
+    },
+    {
+      "epoch": 14.39977973568282,
+      "grad_norm": 0.18345648050308228,
+      "learning_rate": 7.102564102564104e-06,
+      "loss": 0.0089,
+      "step": 13075
+    },
+    {
+      "epoch": 14.427312775330396,
+      "grad_norm": 0.14921247959136963,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 0.0085,
+      "step": 13100
+    },
+    {
+      "epoch": 14.454845814977974,
+      "grad_norm": 0.04616712033748627,
+      "learning_rate": 7.051282051282053e-06,
+      "loss": 0.0083,
+      "step": 13125
+    },
+    {
+      "epoch": 14.482378854625551,
+      "grad_norm": 0.5685108304023743,
+      "learning_rate": 7.025641025641025e-06,
+      "loss": 0.0106,
+      "step": 13150
+    },
+    {
+      "epoch": 14.509911894273127,
+      "grad_norm": 0.13684484362602234,
+      "learning_rate": 7e-06,
+      "loss": 0.0103,
+      "step": 13175
+    },
+    {
+      "epoch": 14.537444933920705,
+      "grad_norm": 0.17398440837860107,
+      "learning_rate": 6.974358974358974e-06,
+      "loss": 0.0095,
+      "step": 13200
+    },
+    {
+      "epoch": 14.564977973568283,
+      "grad_norm": 0.13759827613830566,
+      "learning_rate": 6.948717948717949e-06,
+      "loss": 0.0085,
+      "step": 13225
+    },
+    {
+      "epoch": 14.592511013215859,
+      "grad_norm": 0.06375741213560104,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.0081,
+      "step": 13250
+    },
+    {
+      "epoch": 14.620044052863436,
+      "grad_norm": 0.37382373213768005,
+      "learning_rate": 6.897435897435898e-06,
+      "loss": 0.0084,
+      "step": 13275
+    },
+    {
+      "epoch": 14.647577092511014,
+      "grad_norm": 0.09081553667783737,
+      "learning_rate": 6.871794871794872e-06,
+      "loss": 0.0097,
+      "step": 13300
+    },
+    {
+      "epoch": 14.67511013215859,
+      "grad_norm": 0.06937942653894424,
+      "learning_rate": 6.846153846153847e-06,
+      "loss": 0.01,
+      "step": 13325
+    },
+    {
+      "epoch": 14.702643171806168,
+      "grad_norm": 0.11505573987960815,
+      "learning_rate": 6.820512820512821e-06,
+      "loss": 0.0072,
+      "step": 13350
+    },
+    {
+      "epoch": 14.730176211453745,
+      "grad_norm": 0.2015366107225418,
+      "learning_rate": 6.794871794871796e-06,
+      "loss": 0.0093,
+      "step": 13375
+    },
+    {
+      "epoch": 14.757709251101321,
+      "grad_norm": 0.21869924664497375,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.01,
+      "step": 13400
+    },
+    {
+      "epoch": 14.785242290748899,
+      "grad_norm": 0.1208106279373169,
+      "learning_rate": 6.743589743589745e-06,
+      "loss": 0.009,
+      "step": 13425
+    },
+    {
+      "epoch": 14.812775330396477,
+      "grad_norm": 0.07886633276939392,
+      "learning_rate": 6.717948717948718e-06,
+      "loss": 0.0097,
+      "step": 13450
+    },
+    {
+      "epoch": 14.840308370044053,
+      "grad_norm": 0.26228100061416626,
+      "learning_rate": 6.692307692307692e-06,
+      "loss": 0.0078,
+      "step": 13475
+    },
+    {
+      "epoch": 14.86784140969163,
+      "grad_norm": 0.38411983847618103,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.0098,
+      "step": 13500
+    },
+    {
+      "epoch": 14.895374449339208,
+      "grad_norm": 0.33114126324653625,
+      "learning_rate": 6.641025641025641e-06,
+      "loss": 0.0087,
+      "step": 13525
+    },
+    {
+      "epoch": 14.922907488986784,
+      "grad_norm": 0.10805019736289978,
+      "learning_rate": 6.615384615384616e-06,
+      "loss": 0.007,
+      "step": 13550
+    },
+    {
+      "epoch": 14.950440528634362,
+      "grad_norm": 0.07796984165906906,
+      "learning_rate": 6.58974358974359e-06,
+      "loss": 0.0089,
+      "step": 13575
+    },
+    {
+      "epoch": 14.97797356828194,
+      "grad_norm": 0.18418745696544647,
+      "learning_rate": 6.564102564102565e-06,
+      "loss": 0.0092,
+      "step": 13600
+    },
+    {
+      "epoch": 15.005506607929515,
+      "grad_norm": 0.2542673647403717,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.0079,
+      "step": 13625
+    },
+    {
+      "epoch": 15.033039647577093,
+      "grad_norm": 0.8175550103187561,
+      "learning_rate": 6.512820512820514e-06,
+      "loss": 0.0075,
+      "step": 13650
+    },
+    {
+      "epoch": 15.060572687224669,
+      "grad_norm": 0.2477434277534485,
+      "learning_rate": 6.487179487179488e-06,
+      "loss": 0.0076,
+      "step": 13675
+    },
+    {
+      "epoch": 15.088105726872246,
+      "grad_norm": 0.1775578409433365,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 0.0093,
+      "step": 13700
+    },
+    {
+      "epoch": 15.115638766519824,
+      "grad_norm": 0.043849751353263855,
+      "learning_rate": 6.435897435897437e-06,
+      "loss": 0.0065,
+      "step": 13725
+    },
+    {
+      "epoch": 15.1431718061674,
+      "grad_norm": 0.391318142414093,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 0.0064,
+      "step": 13750
+    },
+    {
+      "epoch": 15.170704845814978,
+      "grad_norm": 0.057629019021987915,
+      "learning_rate": 6.384615384615384e-06,
+      "loss": 0.007,
+      "step": 13775
+    },
+    {
+      "epoch": 15.198237885462555,
+      "grad_norm": 0.3805502951145172,
+      "learning_rate": 6.358974358974359e-06,
+      "loss": 0.0096,
+      "step": 13800
+    },
+    {
+      "epoch": 15.225770925110131,
+      "grad_norm": 0.09331201016902924,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 0.0063,
+      "step": 13825
+    },
+    {
+      "epoch": 15.253303964757709,
+      "grad_norm": 0.26768723130226135,
+      "learning_rate": 6.307692307692308e-06,
+      "loss": 0.0059,
+      "step": 13850
+    },
+    {
+      "epoch": 15.280837004405287,
+      "grad_norm": 0.03280387446284294,
+      "learning_rate": 6.282051282051282e-06,
+      "loss": 0.006,
+      "step": 13875
+    },
+    {
+      "epoch": 15.308370044052863,
+      "grad_norm": 0.050882868468761444,
+      "learning_rate": 6.256410256410257e-06,
+      "loss": 0.0074,
+      "step": 13900
+    },
+    {
+      "epoch": 15.33590308370044,
+      "grad_norm": 0.06220351159572601,
+      "learning_rate": 6.230769230769231e-06,
+      "loss": 0.0061,
+      "step": 13925
+    },
+    {
+      "epoch": 15.363436123348018,
+      "grad_norm": 0.034684717655181885,
+      "learning_rate": 6.205128205128206e-06,
+      "loss": 0.007,
+      "step": 13950
+    },
+    {
+      "epoch": 15.390969162995594,
+      "grad_norm": 0.31570619344711304,
+      "learning_rate": 6.17948717948718e-06,
+      "loss": 0.0074,
+      "step": 13975
+    },
+    {
+      "epoch": 15.418502202643172,
+      "grad_norm": 0.13296958804130554,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.0092,
+      "step": 14000
+    },
+    {
+      "epoch": 15.418502202643172,
+      "eval_cer": 53.11923240058235,
+      "eval_loss": 0.930982768535614,
+      "eval_runtime": 896.4107,
+      "eval_samples_per_second": 11.804,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 94.03111739745404,
+      "step": 14000
+    },
+    {
+      "epoch": 15.44603524229075,
+      "grad_norm": 0.05069255828857422,
+      "learning_rate": 6.128205128205129e-06,
+      "loss": 0.0061,
+      "step": 14025
+    },
+    {
+      "epoch": 15.473568281938325,
+      "grad_norm": 0.30910825729370117,
+      "learning_rate": 6.102564102564104e-06,
+      "loss": 0.0063,
+      "step": 14050
+    },
+    {
+      "epoch": 15.501101321585903,
+      "grad_norm": 0.04979943856596947,
+      "learning_rate": 6.076923076923077e-06,
+      "loss": 0.0062,
+      "step": 14075
+    },
+    {
+      "epoch": 15.52863436123348,
+      "grad_norm": 0.273824006319046,
+      "learning_rate": 6.051282051282051e-06,
+      "loss": 0.0052,
+      "step": 14100
+    },
+    {
+      "epoch": 15.556167400881057,
+      "grad_norm": 0.04723271727561951,
+      "learning_rate": 6.025641025641026e-06,
+      "loss": 0.007,
+      "step": 14125
+    },
+    {
+      "epoch": 15.583700440528634,
+      "grad_norm": 0.08832963556051254,
+      "learning_rate": 6e-06,
+      "loss": 0.0062,
+      "step": 14150
+    },
+    {
+      "epoch": 15.611233480176212,
+      "grad_norm": 0.36821451783180237,
+      "learning_rate": 5.974358974358975e-06,
+      "loss": 0.0066,
+      "step": 14175
+    },
+    {
+      "epoch": 15.638766519823788,
+      "grad_norm": 0.04720625281333923,
+      "learning_rate": 5.948717948717949e-06,
+      "loss": 0.0052,
+      "step": 14200
+    },
+    {
+      "epoch": 15.666299559471366,
+      "grad_norm": 0.05838828161358833,
+      "learning_rate": 5.923076923076924e-06,
+      "loss": 0.0061,
+      "step": 14225
+    },
+    {
+      "epoch": 15.693832599118943,
+      "grad_norm": 0.40729036927223206,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 0.0081,
+      "step": 14250
+    },
+    {
+      "epoch": 15.72136563876652,
+      "grad_norm": 0.04247939959168434,
+      "learning_rate": 5.871794871794873e-06,
+      "loss": 0.007,
+      "step": 14275
+    },
+    {
+      "epoch": 15.748898678414097,
+      "grad_norm": 0.0786793977022171,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 0.0089,
+      "step": 14300
+    },
+    {
+      "epoch": 15.776431718061675,
+      "grad_norm": 0.066864013671875,
+      "learning_rate": 5.820512820512822e-06,
+      "loss": 0.0066,
+      "step": 14325
+    },
+    {
+      "epoch": 15.80396475770925,
+      "grad_norm": 0.05071844533085823,
+      "learning_rate": 5.794871794871796e-06,
+      "loss": 0.0087,
+      "step": 14350
+    },
+    {
+      "epoch": 15.831497797356828,
+      "grad_norm": 0.03305233642458916,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 0.0054,
+      "step": 14375
+    },
+    {
+      "epoch": 15.859030837004406,
+      "grad_norm": 0.1291572004556656,
+      "learning_rate": 5.743589743589743e-06,
+      "loss": 0.0048,
+      "step": 14400
+    },
+    {
+      "epoch": 15.886563876651982,
+      "grad_norm": 0.13923364877700806,
+      "learning_rate": 5.717948717948718e-06,
+      "loss": 0.0059,
+      "step": 14425
+    },
+    {
+      "epoch": 15.91409691629956,
+      "grad_norm": 0.09947941452264786,
+      "learning_rate": 5.692307692307692e-06,
+      "loss": 0.0063,
+      "step": 14450
+    },
+    {
+      "epoch": 15.941629955947137,
+      "grad_norm": 0.27415186166763306,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 0.0088,
+      "step": 14475
+    },
+    {
+      "epoch": 15.969162995594713,
+      "grad_norm": 0.30286744236946106,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.0076,
+      "step": 14500
+    },
+    {
+      "epoch": 15.996696035242291,
+      "grad_norm": 0.07530049234628677,
+      "learning_rate": 5.615384615384616e-06,
+      "loss": 0.0074,
+      "step": 14525
+    },
+    {
+      "epoch": 16.024229074889867,
+      "grad_norm": 0.24431009590625763,
+      "learning_rate": 5.58974358974359e-06,
+      "loss": 0.0056,
+      "step": 14550
+    },
+    {
+      "epoch": 16.051762114537446,
+      "grad_norm": 0.02309373766183853,
+      "learning_rate": 5.564102564102565e-06,
+      "loss": 0.0047,
+      "step": 14575
+    },
+    {
+      "epoch": 16.079295154185022,
+      "grad_norm": 0.02125135436654091,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.0054,
+      "step": 14600
+    },
+    {
+      "epoch": 16.106828193832598,
+      "grad_norm": 0.029789986088871956,
+      "learning_rate": 5.512820512820514e-06,
+      "loss": 0.0039,
+      "step": 14625
+    },
+    {
+      "epoch": 16.134361233480178,
+      "grad_norm": 0.022308455780148506,
+      "learning_rate": 5.487179487179488e-06,
+      "loss": 0.0044,
+      "step": 14650
+    },
+    {
+      "epoch": 16.161894273127754,
+      "grad_norm": 0.05178627371788025,
+      "learning_rate": 5.461538461538461e-06,
+      "loss": 0.0065,
+      "step": 14675
+    },
+    {
+      "epoch": 16.18942731277533,
+      "grad_norm": 0.039131879806518555,
+      "learning_rate": 5.435897435897436e-06,
+      "loss": 0.0042,
+      "step": 14700
+    },
+    {
+      "epoch": 16.21696035242291,
+      "grad_norm": 0.03180893883109093,
+      "learning_rate": 5.41025641025641e-06,
+      "loss": 0.0061,
+      "step": 14725
+    },
+    {
+      "epoch": 16.244493392070485,
+      "grad_norm": 0.2699941098690033,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.0037,
+      "step": 14750
+    },
+    {
+      "epoch": 16.27202643171806,
+      "grad_norm": 0.17495819926261902,
+      "learning_rate": 5.358974358974359e-06,
+      "loss": 0.0056,
+      "step": 14775
+    },
+    {
+      "epoch": 16.29955947136564,
+      "grad_norm": 0.1452856808900833,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.0041,
+      "step": 14800
+    },
+    {
+      "epoch": 16.327092511013216,
+      "grad_norm": 0.042813967913389206,
+      "learning_rate": 5.307692307692308e-06,
+      "loss": 0.0056,
+      "step": 14825
+    },
+    {
+      "epoch": 16.354625550660792,
+      "grad_norm": 0.020087067037820816,
+      "learning_rate": 5.282051282051283e-06,
+      "loss": 0.0039,
+      "step": 14850
+    },
+    {
+      "epoch": 16.38215859030837,
+      "grad_norm": 0.12951348721981049,
+      "learning_rate": 5.256410256410257e-06,
+      "loss": 0.0057,
+      "step": 14875
+    },
+    {
+      "epoch": 16.409691629955947,
+      "grad_norm": 0.13158725202083588,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 0.0049,
+      "step": 14900
+    },
+    {
+      "epoch": 16.437224669603523,
+      "grad_norm": 0.07236882299184799,
+      "learning_rate": 5.205128205128206e-06,
+      "loss": 0.0097,
+      "step": 14925
+    },
+    {
+      "epoch": 16.464757709251103,
+      "grad_norm": 0.2643989324569702,
+      "learning_rate": 5.179487179487181e-06,
+      "loss": 0.0058,
+      "step": 14950
+    },
+    {
+      "epoch": 16.49229074889868,
+      "grad_norm": 0.09582165628671646,
+      "learning_rate": 5.1538461538461534e-06,
+      "loss": 0.0092,
+      "step": 14975
+    },
+    {
+      "epoch": 16.519823788546255,
+      "grad_norm": 0.12667202949523926,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.0069,
+      "step": 15000
+    },
+    {
+      "epoch": 16.519823788546255,
+      "eval_cer": 51.977462657480544,
+      "eval_loss": 0.9370369911193848,
+      "eval_runtime": 863.3538,
+      "eval_samples_per_second": 12.256,
+      "eval_steps_per_second": 3.065,
+      "eval_wer": 93.88024516737387,
+      "step": 15000
+    },
+    {
+      "epoch": 16.547356828193834,
+      "grad_norm": 0.09567159414291382,
+      "learning_rate": 5.1025641025641024e-06,
+      "loss": 0.0038,
+      "step": 15025
+    },
+    {
+      "epoch": 16.57488986784141,
+      "grad_norm": 0.15499436855316162,
+      "learning_rate": 5.076923076923077e-06,
+      "loss": 0.006,
+      "step": 15050
+    },
+    {
+      "epoch": 16.602422907488986,
+      "grad_norm": 0.043445032089948654,
+      "learning_rate": 5.051282051282051e-06,
+      "loss": 0.0044,
+      "step": 15075
+    },
+    {
+      "epoch": 16.629955947136565,
+      "grad_norm": 0.24302434921264648,
+      "learning_rate": 5.025641025641026e-06,
+      "loss": 0.0043,
+      "step": 15100
+    },
+    {
+      "epoch": 16.65748898678414,
+      "grad_norm": 0.06441080570220947,
+      "learning_rate": 5e-06,
+      "loss": 0.0038,
+      "step": 15125
+    },
+    {
+      "epoch": 16.685022026431717,
+      "grad_norm": 0.022898901253938675,
+      "learning_rate": 4.974358974358975e-06,
+      "loss": 0.0035,
+      "step": 15150
+    },
+    {
+      "epoch": 16.712555066079297,
+      "grad_norm": 0.03623131662607193,
+      "learning_rate": 4.948717948717949e-06,
+      "loss": 0.0046,
+      "step": 15175
+    },
+    {
+      "epoch": 16.740088105726873,
+      "grad_norm": 0.023917708545923233,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.0058,
+      "step": 15200
+    },
+    {
+      "epoch": 16.76762114537445,
+      "grad_norm": 0.21040736138820648,
+      "learning_rate": 4.8974358974358975e-06,
+      "loss": 0.005,
+      "step": 15225
+    },
+    {
+      "epoch": 16.795154185022028,
+      "grad_norm": 0.07802154123783112,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 0.0061,
+      "step": 15250
+    },
+    {
+      "epoch": 16.822687224669604,
+      "grad_norm": 0.03765368461608887,
+      "learning_rate": 4.8461538461538465e-06,
+      "loss": 0.0043,
+      "step": 15275
+    },
+    {
+      "epoch": 16.85022026431718,
+      "grad_norm": 0.06764553487300873,
+      "learning_rate": 4.820512820512821e-06,
+      "loss": 0.0041,
+      "step": 15300
+    },
+    {
+      "epoch": 16.87775330396476,
+      "grad_norm": 0.042156316339969635,
+      "learning_rate": 4.7948717948717955e-06,
+      "loss": 0.0057,
+      "step": 15325
+    },
+    {
+      "epoch": 16.905286343612335,
+      "grad_norm": 0.032781120389699936,
+      "learning_rate": 4.76923076923077e-06,
+      "loss": 0.0052,
+      "step": 15350
+    },
+    {
+      "epoch": 16.93281938325991,
+      "grad_norm": 0.05068659782409668,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 0.0044,
+      "step": 15375
+    },
+    {
+      "epoch": 16.96035242290749,
+      "grad_norm": 0.03536130487918854,
+      "learning_rate": 4.717948717948718e-06,
+      "loss": 0.0053,
+      "step": 15400
+    },
+    {
+      "epoch": 16.987885462555067,
+      "grad_norm": 0.21238285303115845,
+      "learning_rate": 4.692307692307693e-06,
+      "loss": 0.005,
+      "step": 15425
+    },
+    {
+      "epoch": 17.015418502202643,
+      "grad_norm": 0.06675603240728378,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.0044,
+      "step": 15450
+    },
+    {
+      "epoch": 17.042951541850222,
+      "grad_norm": 0.13563722372055054,
+      "learning_rate": 4.641025641025642e-06,
+      "loss": 0.0039,
+      "step": 15475
+    },
+    {
+      "epoch": 17.070484581497798,
+      "grad_norm": 0.04386042803525925,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.0028,
+      "step": 15500
+    },
+    {
+      "epoch": 17.098017621145374,
+      "grad_norm": 0.014782898128032684,
+      "learning_rate": 4.58974358974359e-06,
+      "loss": 0.0029,
+      "step": 15525
+    },
+    {
+      "epoch": 17.125550660792953,
+      "grad_norm": 0.01799604296684265,
+      "learning_rate": 4.564102564102564e-06,
+      "loss": 0.0035,
+      "step": 15550
+    },
+    {
+      "epoch": 17.15308370044053,
+      "grad_norm": 0.03504837304353714,
+      "learning_rate": 4.538461538461539e-06,
+      "loss": 0.0036,
+      "step": 15575
+    },
+    {
+      "epoch": 17.180616740088105,
+      "grad_norm": 0.017535801976919174,
+      "learning_rate": 4.512820512820513e-06,
+      "loss": 0.0021,
+      "step": 15600
+    },
+    {
+      "epoch": 17.208149779735685,
+      "grad_norm": 0.0199030339717865,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 0.0037,
+      "step": 15625
+    },
+    {
+      "epoch": 17.23568281938326,
+      "grad_norm": 0.026048993691802025,
+      "learning_rate": 4.461538461538462e-06,
+      "loss": 0.0049,
+      "step": 15650
+    },
+    {
+      "epoch": 17.263215859030836,
+      "grad_norm": 0.029507996514439583,
+      "learning_rate": 4.435897435897436e-06,
+      "loss": 0.0037,
+      "step": 15675
+    },
+    {
+      "epoch": 17.290748898678412,
+      "grad_norm": 0.25132548809051514,
+      "learning_rate": 4.4102564102564104e-06,
+      "loss": 0.0035,
+      "step": 15700
+    },
+    {
+      "epoch": 17.318281938325992,
+      "grad_norm": 0.02435138076543808,
+      "learning_rate": 4.384615384615385e-06,
+      "loss": 0.0034,
+      "step": 15725
+    },
+    {
+      "epoch": 17.345814977973568,
+      "grad_norm": 0.020107686519622803,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 0.0037,
+      "step": 15750
+    },
+    {
+      "epoch": 17.373348017621144,
+      "grad_norm": 0.07011119276285172,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 0.0024,
+      "step": 15775
+    },
+    {
+      "epoch": 17.400881057268723,
+      "grad_norm": 0.016648728400468826,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 0.0025,
+      "step": 15800
+    },
+    {
+      "epoch": 17.4284140969163,
+      "grad_norm": 0.03225812688469887,
+      "learning_rate": 4.282051282051282e-06,
+      "loss": 0.0043,
+      "step": 15825
+    },
+    {
+      "epoch": 17.455947136563875,
+      "grad_norm": 0.028381196781992912,
+      "learning_rate": 4.2564102564102566e-06,
+      "loss": 0.0044,
+      "step": 15850
+    },
+    {
+      "epoch": 17.483480176211454,
+      "grad_norm": 0.01819649152457714,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 0.0032,
+      "step": 15875
+    },
+    {
+      "epoch": 17.51101321585903,
+      "grad_norm": 0.018908126279711723,
+      "learning_rate": 4.2051282051282055e-06,
+      "loss": 0.0047,
+      "step": 15900
+    },
+    {
+      "epoch": 17.538546255506606,
+      "grad_norm": 0.022597506642341614,
+      "learning_rate": 4.17948717948718e-06,
+      "loss": 0.0039,
+      "step": 15925
+    },
+    {
+      "epoch": 17.566079295154186,
+      "grad_norm": 0.024837888777256012,
+      "learning_rate": 4.1538461538461545e-06,
+      "loss": 0.0031,
+      "step": 15950
+    },
+    {
+      "epoch": 17.59361233480176,
+      "grad_norm": 0.025047263130545616,
+      "learning_rate": 4.128205128205128e-06,
+      "loss": 0.0028,
+      "step": 15975
+    },
+    {
+      "epoch": 17.621145374449338,
+      "grad_norm": 0.021962720900774002,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.0023,
+      "step": 16000
+    },
+    {
+      "epoch": 17.621145374449338,
+      "eval_cer": 49.68761821787626,
+      "eval_loss": 0.9436636567115784,
+      "eval_runtime": 856.2417,
+      "eval_samples_per_second": 12.357,
+      "eval_steps_per_second": 3.09,
+      "eval_wer": 94.23856671381424,
+      "step": 16000
+    },
+    {
+      "epoch": 17.648678414096917,
+      "grad_norm": 0.024941807612776756,
+      "learning_rate": 4.076923076923077e-06,
+      "loss": 0.0032,
+      "step": 16025
+    },
+    {
+      "epoch": 17.676211453744493,
+      "grad_norm": 0.018255887553095818,
+      "learning_rate": 4.051282051282052e-06,
+      "loss": 0.0055,
+      "step": 16050
+    },
+    {
+      "epoch": 17.70374449339207,
+      "grad_norm": 0.08047758787870407,
+      "learning_rate": 4.025641025641026e-06,
+      "loss": 0.003,
+      "step": 16075
+    },
+    {
+      "epoch": 17.73127753303965,
+      "grad_norm": 0.028241898864507675,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0036,
+      "step": 16100
+    },
+    {
+      "epoch": 17.758810572687224,
+      "grad_norm": 0.12309114634990692,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 0.0036,
+      "step": 16125
+    },
+    {
+      "epoch": 17.7863436123348,
+      "grad_norm": 0.019569188356399536,
+      "learning_rate": 3.948717948717949e-06,
+      "loss": 0.0039,
+      "step": 16150
+    },
+    {
+      "epoch": 17.81387665198238,
+      "grad_norm": 0.027806563302874565,
+      "learning_rate": 3.923076923076923e-06,
+      "loss": 0.0035,
+      "step": 16175
+    },
+    {
+      "epoch": 17.841409691629956,
+      "grad_norm": 0.05405218526721001,
+      "learning_rate": 3.897435897435898e-06,
+      "loss": 0.0036,
+      "step": 16200
+    },
+    {
+      "epoch": 17.86894273127753,
+      "grad_norm": 0.15282440185546875,
+      "learning_rate": 3.871794871794872e-06,
+      "loss": 0.0037,
+      "step": 16225
+    },
+    {
+      "epoch": 17.89647577092511,
+      "grad_norm": 0.023264387622475624,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.0028,
+      "step": 16250
+    },
+    {
+      "epoch": 17.924008810572687,
+      "grad_norm": 0.07771216332912445,
+      "learning_rate": 3.8205128205128204e-06,
+      "loss": 0.0065,
+      "step": 16275
+    },
+    {
+      "epoch": 17.951541850220263,
+      "grad_norm": 0.289384663105011,
+      "learning_rate": 3.794871794871795e-06,
+      "loss": 0.0085,
+      "step": 16300
+    },
+    {
+      "epoch": 17.979074889867842,
+      "grad_norm": 0.050969693809747696,
+      "learning_rate": 3.7692307692307694e-06,
+      "loss": 0.0048,
+      "step": 16325
+    },
+    {
+      "epoch": 18.006607929515418,
+      "grad_norm": 0.029267285019159317,
+      "learning_rate": 3.743589743589744e-06,
+      "loss": 0.0029,
+      "step": 16350
+    },
+    {
+      "epoch": 18.034140969162994,
+      "grad_norm": 0.04442453011870384,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 0.0042,
+      "step": 16375
+    },
+    {
+      "epoch": 18.061674008810574,
+      "grad_norm": 0.019591569900512695,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 0.0035,
+      "step": 16400
+    },
+    {
+      "epoch": 18.08920704845815,
+      "grad_norm": 0.041334908455610275,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 0.005,
+      "step": 16425
+    },
+    {
+      "epoch": 18.116740088105725,
+      "grad_norm": 0.04090972989797592,
+      "learning_rate": 3.641025641025641e-06,
+      "loss": 0.0045,
+      "step": 16450
+    },
+    {
+      "epoch": 18.144273127753305,
+      "grad_norm": 0.024983281269669533,
+      "learning_rate": 3.6153846153846156e-06,
+      "loss": 0.0028,
+      "step": 16475
+    },
+    {
+      "epoch": 18.17180616740088,
+      "grad_norm": 0.0352012999355793,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 0.0026,
+      "step": 16500
+    },
+    {
+      "epoch": 18.199339207048457,
+      "grad_norm": 0.017669979482889175,
+      "learning_rate": 3.5641025641025646e-06,
+      "loss": 0.0028,
+      "step": 16525
+    },
+    {
+      "epoch": 18.226872246696036,
+      "grad_norm": 0.015492716804146767,
+      "learning_rate": 3.538461538461539e-06,
+      "loss": 0.0024,
+      "step": 16550
+    },
+    {
+      "epoch": 18.254405286343612,
+      "grad_norm": 0.031913984566926956,
+      "learning_rate": 3.5128205128205127e-06,
+      "loss": 0.0029,
+      "step": 16575
+    },
+    {
+      "epoch": 18.281938325991188,
+      "grad_norm": 0.019828276708722115,
+      "learning_rate": 3.487179487179487e-06,
+      "loss": 0.0028,
+      "step": 16600
+    },
+    {
+      "epoch": 18.309471365638768,
+      "grad_norm": 0.015073757618665695,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.002,
+      "step": 16625
+    },
+    {
+      "epoch": 18.337004405286343,
+      "grad_norm": 0.03369276598095894,
+      "learning_rate": 3.435897435897436e-06,
+      "loss": 0.0023,
+      "step": 16650
+    },
+    {
+      "epoch": 18.36453744493392,
+      "grad_norm": 0.028322968631982803,
+      "learning_rate": 3.4102564102564107e-06,
+      "loss": 0.0039,
+      "step": 16675
+    },
+    {
+      "epoch": 18.3920704845815,
+      "grad_norm": 0.03424219414591789,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 0.0028,
+      "step": 16700
+    },
+    {
+      "epoch": 18.419603524229075,
+      "grad_norm": 0.04083619639277458,
+      "learning_rate": 3.358974358974359e-06,
+      "loss": 0.0026,
+      "step": 16725
+    },
+    {
+      "epoch": 18.44713656387665,
+      "grad_norm": 0.023958973586559296,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.0033,
+      "step": 16750
+    },
+    {
+      "epoch": 18.47466960352423,
+      "grad_norm": 0.03387632593512535,
+      "learning_rate": 3.307692307692308e-06,
+      "loss": 0.0024,
+      "step": 16775
+    },
+    {
+      "epoch": 18.502202643171806,
+      "grad_norm": 0.033548954874277115,
+      "learning_rate": 3.2820512820512823e-06,
+      "loss": 0.0046,
+      "step": 16800
+    },
+    {
+      "epoch": 18.529735682819382,
+      "grad_norm": 0.020479055121541023,
+      "learning_rate": 3.256410256410257e-06,
+      "loss": 0.0027,
+      "step": 16825
+    },
+    {
+      "epoch": 18.55726872246696,
+      "grad_norm": 0.07224612683057785,
+      "learning_rate": 3.2307692307692313e-06,
+      "loss": 0.0038,
+      "step": 16850
+    },
+    {
+      "epoch": 18.584801762114537,
+      "grad_norm": 0.02132277749478817,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 0.0028,
+      "step": 16875
+    },
+    {
+      "epoch": 18.612334801762113,
+      "grad_norm": 0.014213715679943562,
+      "learning_rate": 3.1794871794871795e-06,
+      "loss": 0.0021,
+      "step": 16900
+    },
+    {
+      "epoch": 18.639867841409693,
+      "grad_norm": 0.11814253032207489,
+      "learning_rate": 3.153846153846154e-06,
+      "loss": 0.0031,
+      "step": 16925
+    },
+    {
+      "epoch": 18.66740088105727,
+      "grad_norm": 0.11767589300870895,
+      "learning_rate": 3.1282051282051284e-06,
+      "loss": 0.0039,
+      "step": 16950
+    },
+    {
+      "epoch": 18.694933920704845,
+      "grad_norm": 0.01782354526221752,
+      "learning_rate": 3.102564102564103e-06,
+      "loss": 0.0029,
+      "step": 16975
+    },
+    {
+      "epoch": 18.722466960352424,
+      "grad_norm": 0.22178053855895996,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.0026,
+      "step": 17000
+    },
+    {
+      "epoch": 18.722466960352424,
+      "eval_cer": 51.592860500040125,
+      "eval_loss": 0.9495370984077454,
+      "eval_runtime": 876.7679,
+      "eval_samples_per_second": 12.068,
+      "eval_steps_per_second": 3.018,
+      "eval_wer": 94.0971239981141,
+      "step": 17000
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 0.028327735140919685,
+      "learning_rate": 3.051282051282052e-06,
+      "loss": 0.0024,
+      "step": 17025
+    },
+    {
+      "epoch": 18.777533039647576,
+      "grad_norm": 0.2968517243862152,
+      "learning_rate": 3.0256410256410256e-06,
+      "loss": 0.002,
+      "step": 17050
+    },
+    {
+      "epoch": 18.805066079295155,
+      "grad_norm": 0.02121434547007084,
+      "learning_rate": 3e-06,
+      "loss": 0.003,
+      "step": 17075
+    },
+    {
+      "epoch": 18.83259911894273,
+      "grad_norm": 0.02059946022927761,
+      "learning_rate": 2.9743589743589746e-06,
+      "loss": 0.0022,
+      "step": 17100
+    },
+    {
+      "epoch": 18.860132158590307,
+      "grad_norm": 0.03535676375031471,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 0.0028,
+      "step": 17125
+    },
+    {
+      "epoch": 18.887665198237887,
+      "grad_norm": 0.017292790114879608,
+      "learning_rate": 2.9230769230769236e-06,
+      "loss": 0.0032,
+      "step": 17150
+    },
+    {
+      "epoch": 18.915198237885463,
+      "grad_norm": 0.010941505432128906,
+      "learning_rate": 2.897435897435898e-06,
+      "loss": 0.002,
+      "step": 17175
+    },
+    {
+      "epoch": 18.94273127753304,
+      "grad_norm": 0.07651369273662567,
+      "learning_rate": 2.8717948717948717e-06,
+      "loss": 0.0024,
+      "step": 17200
+    },
+    {
+      "epoch": 18.970264317180618,
+      "grad_norm": 0.033956125378608704,
+      "learning_rate": 2.846153846153846e-06,
+      "loss": 0.0021,
+      "step": 17225
+    },
+    {
+      "epoch": 18.997797356828194,
+      "grad_norm": 0.01717967540025711,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 0.0039,
+      "step": 17250
+    },
+    {
+      "epoch": 19.02533039647577,
+      "grad_norm": 0.06303483992815018,
+      "learning_rate": 2.794871794871795e-06,
+      "loss": 0.0026,
+      "step": 17275
+    },
+    {
+      "epoch": 19.05286343612335,
+      "grad_norm": 0.024857090786099434,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 0.0018,
+      "step": 17300
+    },
+    {
+      "epoch": 19.080396475770925,
+      "grad_norm": 0.02651827596127987,
+      "learning_rate": 2.743589743589744e-06,
+      "loss": 0.0018,
+      "step": 17325
+    },
+    {
+      "epoch": 19.1079295154185,
+      "grad_norm": 0.018129022791981697,
+      "learning_rate": 2.717948717948718e-06,
+      "loss": 0.0016,
+      "step": 17350
+    },
+    {
+      "epoch": 19.13546255506608,
+      "grad_norm": 0.012260057963430882,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 0.0023,
+      "step": 17375
+    },
+    {
+      "epoch": 19.162995594713657,
+      "grad_norm": 0.013982090167701244,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.0016,
+      "step": 17400
+    },
+    {
+      "epoch": 19.190528634361232,
+      "grad_norm": 0.02801360934972763,
+      "learning_rate": 2.6410256410256413e-06,
+      "loss": 0.0019,
+      "step": 17425
+    },
+    {
+      "epoch": 19.218061674008812,
+      "grad_norm": 0.016421474516391754,
+      "learning_rate": 2.615384615384616e-06,
+      "loss": 0.0022,
+      "step": 17450
+    },
+    {
+      "epoch": 19.245594713656388,
+      "grad_norm": 0.27571967244148254,
+      "learning_rate": 2.5897435897435903e-06,
+      "loss": 0.0025,
+      "step": 17475
+    },
+    {
+      "epoch": 19.273127753303964,
+      "grad_norm": 0.0503142848610878,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.0017,
+      "step": 17500
+    },
+    {
+      "epoch": 19.300660792951543,
+      "grad_norm": 0.02715384215116501,
+      "learning_rate": 2.5384615384615385e-06,
+      "loss": 0.002,
+      "step": 17525
+    },
+    {
+      "epoch": 19.32819383259912,
+      "grad_norm": 0.01945788972079754,
+      "learning_rate": 2.512820512820513e-06,
+      "loss": 0.0018,
+      "step": 17550
+    },
+    {
+      "epoch": 19.355726872246695,
+      "grad_norm": 0.010799442417919636,
+      "learning_rate": 2.4871794871794875e-06,
+      "loss": 0.0016,
+      "step": 17575
+    },
+    {
+      "epoch": 19.383259911894275,
+      "grad_norm": 0.02447289600968361,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 0.0026,
+      "step": 17600
+    },
+    {
+      "epoch": 19.41079295154185,
+      "grad_norm": 0.009756376035511494,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 0.0021,
+      "step": 17625
+    },
+    {
+      "epoch": 19.438325991189426,
+      "grad_norm": 0.013634881004691124,
+      "learning_rate": 2.4102564102564105e-06,
+      "loss": 0.0017,
+      "step": 17650
+    },
+    {
+      "epoch": 19.465859030837006,
+      "grad_norm": 0.016096901148557663,
+      "learning_rate": 2.384615384615385e-06,
+      "loss": 0.0025,
+      "step": 17675
+    },
+    {
+      "epoch": 19.493392070484582,
+      "grad_norm": 0.010127868503332138,
+      "learning_rate": 2.358974358974359e-06,
+      "loss": 0.0019,
+      "step": 17700
+    },
+    {
+      "epoch": 19.520925110132158,
+      "grad_norm": 0.026323504745960236,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 0.0026,
+      "step": 17725
+    },
+    {
+      "epoch": 19.548458149779737,
+      "grad_norm": 0.01952183246612549,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.0023,
+      "step": 17750
+    },
+    {
+      "epoch": 19.575991189427313,
+      "grad_norm": 0.013677220791578293,
+      "learning_rate": 2.282051282051282e-06,
+      "loss": 0.0018,
+      "step": 17775
+    },
+    {
+      "epoch": 19.60352422907489,
+      "grad_norm": 0.01971456967294216,
+      "learning_rate": 2.2564102564102566e-06,
+      "loss": 0.0032,
+      "step": 17800
+    },
+    {
+      "epoch": 19.63105726872247,
+      "grad_norm": 0.016475846990942955,
+      "learning_rate": 2.230769230769231e-06,
+      "loss": 0.0021,
+      "step": 17825
+    },
+    {
+      "epoch": 19.658590308370044,
+      "grad_norm": 0.01166547928005457,
+      "learning_rate": 2.2051282051282052e-06,
+      "loss": 0.0015,
+      "step": 17850
+    },
+    {
+      "epoch": 19.68612334801762,
+      "grad_norm": 0.010217426344752312,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 0.0017,
+      "step": 17875
+    },
+    {
+      "epoch": 19.7136563876652,
+      "grad_norm": 0.011065994389355183,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 0.0019,
+      "step": 17900
+    },
+    {
+      "epoch": 19.741189427312776,
+      "grad_norm": 0.011596587486565113,
+      "learning_rate": 2.1282051282051283e-06,
+      "loss": 0.0016,
+      "step": 17925
+    },
+    {
+      "epoch": 19.76872246696035,
+      "grad_norm": 0.010966944508254528,
+      "learning_rate": 2.1025641025641028e-06,
+      "loss": 0.0028,
+      "step": 17950
+    },
+    {
+      "epoch": 19.79625550660793,
+      "grad_norm": 0.014122538268566132,
+      "learning_rate": 2.0769230769230773e-06,
+      "loss": 0.0025,
+      "step": 17975
+    },
+    {
+      "epoch": 19.823788546255507,
+      "grad_norm": 0.01831120438873768,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 0.0016,
+      "step": 18000
+    },
+    {
+      "epoch": 19.823788546255507,
+      "eval_cer": 51.094196003805905,
+      "eval_loss": 0.9530871510505676,
+      "eval_runtime": 859.385,
+      "eval_samples_per_second": 12.312,
+      "eval_steps_per_second": 3.079,
+      "eval_wer": 94.16313059877416,
+      "step": 18000
+    },
+    {
+      "epoch": 19.851321585903083,
+      "grad_norm": 0.011395452544093132,
+      "learning_rate": 2.025641025641026e-06,
+      "loss": 0.0019,
+      "step": 18025
+    },
+    {
+      "epoch": 19.878854625550662,
+      "grad_norm": 0.009908227249979973,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.0016,
+      "step": 18050
+    },
+    {
+      "epoch": 19.90638766519824,
+      "grad_norm": 0.015528595075011253,
+      "learning_rate": 1.9743589743589744e-06,
+      "loss": 0.0027,
+      "step": 18075
+    },
+    {
+      "epoch": 19.933920704845814,
+      "grad_norm": 0.010827014222741127,
+      "learning_rate": 1.948717948717949e-06,
+      "loss": 0.0025,
+      "step": 18100
+    },
+    {
+      "epoch": 19.961453744493394,
+      "grad_norm": 0.012095007114112377,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.0016,
+      "step": 18125
+    },
+    {
+      "epoch": 19.98898678414097,
+      "grad_norm": 0.0175089742988348,
+      "learning_rate": 1.8974358974358975e-06,
+      "loss": 0.0026,
+      "step": 18150
+    },
+    {
+      "epoch": 20.016519823788546,
+      "grad_norm": 0.008043349720537663,
+      "learning_rate": 1.871794871794872e-06,
+      "loss": 0.0016,
+      "step": 18175
+    },
+    {
+      "epoch": 20.044052863436125,
+      "grad_norm": 0.012720318511128426,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 0.0014,
+      "step": 18200
+    },
+    {
+      "epoch": 20.0715859030837,
+      "grad_norm": 0.008536312729120255,
+      "learning_rate": 1.8205128205128205e-06,
+      "loss": 0.0015,
+      "step": 18225
+    },
+    {
+      "epoch": 20.099118942731277,
+      "grad_norm": 0.007359918672591448,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 0.0016,
+      "step": 18250
+    },
+    {
+      "epoch": 20.126651982378856,
+      "grad_norm": 0.007894999347627163,
+      "learning_rate": 1.7692307692307695e-06,
+      "loss": 0.0013,
+      "step": 18275
+    },
+    {
+      "epoch": 20.154185022026432,
+      "grad_norm": 0.011457731947302818,
+      "learning_rate": 1.7435897435897436e-06,
+      "loss": 0.0013,
+      "step": 18300
+    },
+    {
+      "epoch": 20.181718061674008,
+      "grad_norm": 0.013349108397960663,
+      "learning_rate": 1.717948717948718e-06,
+      "loss": 0.0014,
+      "step": 18325
+    },
+    {
+      "epoch": 20.209251101321588,
+      "grad_norm": 0.009640936739742756,
+      "learning_rate": 1.6923076923076926e-06,
+      "loss": 0.0015,
+      "step": 18350
+    },
+    {
+      "epoch": 20.236784140969164,
+      "grad_norm": 0.009478888474404812,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.002,
+      "step": 18375
+    },
+    {
+      "epoch": 20.26431718061674,
+      "grad_norm": 0.010656708851456642,
+      "learning_rate": 1.6410256410256412e-06,
+      "loss": 0.0014,
+      "step": 18400
+    },
+    {
+      "epoch": 20.291850220264315,
+      "grad_norm": 0.009404104202985764,
+      "learning_rate": 1.6153846153846157e-06,
+      "loss": 0.0014,
+      "step": 18425
+    },
+    {
+      "epoch": 20.319383259911895,
+      "grad_norm": 0.0218101404607296,
+      "learning_rate": 1.5897435897435897e-06,
+      "loss": 0.0017,
+      "step": 18450
+    },
+    {
+      "epoch": 20.34691629955947,
+      "grad_norm": 0.007058306131511927,
+      "learning_rate": 1.5641025641025642e-06,
+      "loss": 0.0013,
+      "step": 18475
+    },
+    {
+      "epoch": 20.374449339207047,
+      "grad_norm": 0.012106086127460003,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.0023,
+      "step": 18500
+    },
+    {
+      "epoch": 20.401982378854626,
+      "grad_norm": 0.00871138833463192,
+      "learning_rate": 1.5128205128205128e-06,
+      "loss": 0.0017,
+      "step": 18525
+    },
+    {
+      "epoch": 20.429515418502202,
+      "grad_norm": 0.012688432820141315,
+      "learning_rate": 1.4871794871794873e-06,
+      "loss": 0.0015,
+      "step": 18550
+    },
+    {
+      "epoch": 20.457048458149778,
+      "grad_norm": 0.008517405949532986,
+      "learning_rate": 1.4615384615384618e-06,
+      "loss": 0.0014,
+      "step": 18575
+    },
+    {
+      "epoch": 20.484581497797357,
+      "grad_norm": 0.009141705930233002,
+      "learning_rate": 1.4358974358974359e-06,
+      "loss": 0.0012,
+      "step": 18600
+    },
+    {
+      "epoch": 20.512114537444933,
+      "grad_norm": 0.01858842372894287,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 0.0016,
+      "step": 18625
+    },
+    {
+      "epoch": 20.53964757709251,
+      "grad_norm": 0.10274213552474976,
+      "learning_rate": 1.3846153846153848e-06,
+      "loss": 0.0017,
+      "step": 18650
+    },
+    {
+      "epoch": 20.56718061674009,
+      "grad_norm": 0.009239338338375092,
+      "learning_rate": 1.358974358974359e-06,
+      "loss": 0.0014,
+      "step": 18675
+    },
+    {
+      "epoch": 20.594713656387665,
+      "grad_norm": 0.0146435322239995,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.0013,
+      "step": 18700
+    },
+    {
+      "epoch": 20.62224669603524,
+      "grad_norm": 0.01037636585533619,
+      "learning_rate": 1.307692307692308e-06,
+      "loss": 0.0015,
+      "step": 18725
+    },
+    {
+      "epoch": 20.64977973568282,
+      "grad_norm": 0.007970661856234074,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.0012,
+      "step": 18750
+    },
+    {
+      "epoch": 20.677312775330396,
+      "grad_norm": 0.01924486644566059,
+      "learning_rate": 1.2564102564102565e-06,
+      "loss": 0.0016,
+      "step": 18775
+    },
+    {
+      "epoch": 20.704845814977972,
+      "grad_norm": 0.009840236976742744,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 0.0015,
+      "step": 18800
+    },
+    {
+      "epoch": 20.73237885462555,
+      "grad_norm": 0.09442971646785736,
+      "learning_rate": 1.2051282051282053e-06,
+      "loss": 0.0016,
+      "step": 18825
+    },
+    {
+      "epoch": 20.759911894273127,
+      "grad_norm": 0.008129543624818325,
+      "learning_rate": 1.1794871794871795e-06,
+      "loss": 0.0014,
+      "step": 18850
+    },
+    {
+      "epoch": 20.787444933920703,
+      "grad_norm": 0.00982487853616476,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 0.0016,
+      "step": 18875
+    },
+    {
+      "epoch": 20.814977973568283,
+      "grad_norm": 0.009310917928814888,
+      "learning_rate": 1.1282051282051283e-06,
+      "loss": 0.002,
+      "step": 18900
+    },
+    {
+      "epoch": 20.84251101321586,
+      "grad_norm": 0.012313717044889927,
+      "learning_rate": 1.1025641025641026e-06,
+      "loss": 0.0015,
+      "step": 18925
+    },
+    {
+      "epoch": 20.870044052863435,
+      "grad_norm": 0.013325495645403862,
+      "learning_rate": 1.076923076923077e-06,
+      "loss": 0.002,
+      "step": 18950
+    },
+    {
+      "epoch": 20.897577092511014,
+      "grad_norm": 0.009593603201210499,
+      "learning_rate": 1.0512820512820514e-06,
+      "loss": 0.0015,
+      "step": 18975
+    },
+    {
+      "epoch": 20.92511013215859,
+      "grad_norm": 0.013929825276136398,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 0.0015,
+      "step": 19000
+    },
+    {
+      "epoch": 20.92511013215859,
+      "eval_cer": 50.91249871035044,
+      "eval_loss": 0.9608182907104492,
+      "eval_runtime": 847.4888,
+      "eval_samples_per_second": 12.485,
+      "eval_steps_per_second": 3.122,
+      "eval_wer": 94.7006129184347,
+      "step": 19000
+    },
+    {
+      "epoch": 20.952643171806166,
+      "grad_norm": 0.012814809568226337,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.0024,
+      "step": 19025
+    },
+    {
+      "epoch": 20.980176211453745,
+      "grad_norm": 0.009638777002692223,
+      "learning_rate": 9.743589743589745e-07,
+      "loss": 0.0014,
+      "step": 19050
+    },
+    {
+      "epoch": 21.00770925110132,
+      "grad_norm": 0.01127631589770317,
+      "learning_rate": 9.487179487179487e-07,
+      "loss": 0.0014,
+      "step": 19075
+    },
+    {
+      "epoch": 21.035242290748897,
+      "grad_norm": 0.008207294158637524,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 0.002,
+      "step": 19100
+    },
+    {
+      "epoch": 21.062775330396477,
+      "grad_norm": 0.009193181060254574,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 0.0015,
+      "step": 19125
+    },
+    {
+      "epoch": 21.090308370044053,
+      "grad_norm": 0.01055130921304226,
+      "learning_rate": 8.717948717948718e-07,
+      "loss": 0.0013,
+      "step": 19150
+    },
+    {
+      "epoch": 21.11784140969163,
+      "grad_norm": 0.00705339340493083,
+      "learning_rate": 8.461538461538463e-07,
+      "loss": 0.0012,
+      "step": 19175
+    },
+    {
+      "epoch": 21.145374449339208,
+      "grad_norm": 0.012715994380414486,
+      "learning_rate": 8.205128205128206e-07,
+      "loss": 0.0012,
+      "step": 19200
+    },
+    {
+      "epoch": 21.172907488986784,
+      "grad_norm": 0.006566143594682217,
+      "learning_rate": 7.948717948717949e-07,
+      "loss": 0.0013,
+      "step": 19225
+    },
+    {
+      "epoch": 21.20044052863436,
+      "grad_norm": 0.00661145756021142,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 0.0012,
+      "step": 19250
+    },
+    {
+      "epoch": 21.22797356828194,
+      "grad_norm": 0.008768678642809391,
+      "learning_rate": 7.435897435897436e-07,
+      "loss": 0.0013,
+      "step": 19275
+    },
+    {
+      "epoch": 21.255506607929515,
+      "grad_norm": 0.008478008210659027,
+      "learning_rate": 7.179487179487179e-07,
+      "loss": 0.0011,
+      "step": 19300
+    },
+    {
+      "epoch": 21.28303964757709,
+      "grad_norm": 0.012739352881908417,
+      "learning_rate": 6.923076923076924e-07,
+      "loss": 0.0016,
+      "step": 19325
+    },
+    {
+      "epoch": 21.31057268722467,
+      "grad_norm": 0.006703950930386782,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.0012,
+      "step": 19350
+    },
+    {
+      "epoch": 21.338105726872246,
+      "grad_norm": 0.00934862531721592,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 0.0013,
+      "step": 19375
+    },
+    {
+      "epoch": 21.365638766519822,
+      "grad_norm": 0.011370371095836163,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 0.0013,
+      "step": 19400
+    },
+    {
+      "epoch": 21.393171806167402,
+      "grad_norm": 0.010093637742102146,
+      "learning_rate": 5.897435897435898e-07,
+      "loss": 0.0013,
+      "step": 19425
+    },
+    {
+      "epoch": 21.420704845814978,
+      "grad_norm": 0.007553795352578163,
+      "learning_rate": 5.641025641025642e-07,
+      "loss": 0.0013,
+      "step": 19450
+    },
+    {
+      "epoch": 21.448237885462554,
+      "grad_norm": 0.009227064438164234,
+      "learning_rate": 5.384615384615386e-07,
+      "loss": 0.0013,
+      "step": 19475
+    },
+    {
+      "epoch": 21.475770925110133,
+      "grad_norm": 0.006895294412970543,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 0.0013,
+      "step": 19500
+    },
+    {
+      "epoch": 21.50330396475771,
+      "grad_norm": 0.006935523357242346,
+      "learning_rate": 4.871794871794872e-07,
+      "loss": 0.0012,
+      "step": 19525
+    },
+    {
+      "epoch": 21.530837004405285,
+      "grad_norm": 0.009774105623364449,
+      "learning_rate": 4.615384615384616e-07,
+      "loss": 0.0011,
+      "step": 19550
+    },
+    {
+      "epoch": 21.558370044052865,
+      "grad_norm": 0.008834905922412872,
+      "learning_rate": 4.358974358974359e-07,
+      "loss": 0.0011,
+      "step": 19575
+    },
+    {
+      "epoch": 21.58590308370044,
+      "grad_norm": 0.011201854795217514,
+      "learning_rate": 4.102564102564103e-07,
+      "loss": 0.0012,
+      "step": 19600
+    },
+    {
+      "epoch": 21.613436123348016,
+      "grad_norm": 0.006510408595204353,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 0.0013,
+      "step": 19625
+    },
+    {
+      "epoch": 21.640969162995596,
+      "grad_norm": 0.0077276709489524364,
+      "learning_rate": 3.5897435897435896e-07,
+      "loss": 0.0012,
+      "step": 19650
+    },
+    {
+      "epoch": 21.66850220264317,
+      "grad_norm": 0.006777653470635414,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 0.0012,
+      "step": 19675
+    },
+    {
+      "epoch": 21.696035242290748,
+      "grad_norm": 0.007512731943279505,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 0.0013,
+      "step": 19700
+    },
+    {
+      "epoch": 21.723568281938327,
+      "grad_norm": 0.007246058434247971,
+      "learning_rate": 2.820512820512821e-07,
+      "loss": 0.0015,
+      "step": 19725
+    },
+    {
+      "epoch": 21.751101321585903,
+      "grad_norm": 0.007962902076542377,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 0.0013,
+      "step": 19750
+    },
+    {
+      "epoch": 21.77863436123348,
+      "grad_norm": 0.009491794742643833,
+      "learning_rate": 2.307692307692308e-07,
+      "loss": 0.0012,
+      "step": 19775
+    },
+    {
+      "epoch": 21.80616740088106,
+      "grad_norm": 0.012724686414003372,
+      "learning_rate": 2.0512820512820514e-07,
+      "loss": 0.0013,
+      "step": 19800
+    },
+    {
+      "epoch": 21.833700440528634,
+      "grad_norm": 0.0072784461081027985,
+      "learning_rate": 1.7948717948717948e-07,
+      "loss": 0.0012,
+      "step": 19825
+    },
+    {
+      "epoch": 21.86123348017621,
+      "grad_norm": 0.012166726402938366,
+      "learning_rate": 1.5384615384615387e-07,
+      "loss": 0.0012,
+      "step": 19850
+    },
+    {
+      "epoch": 21.88876651982379,
+      "grad_norm": 0.10767704248428345,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 0.0014,
+      "step": 19875
+    },
+    {
+      "epoch": 21.916299559471366,
+      "grad_norm": 0.008811806328594685,
+      "learning_rate": 1.0256410256410257e-07,
+      "loss": 0.0011,
+      "step": 19900
+    },
+    {
+      "epoch": 21.94383259911894,
+      "grad_norm": 0.010796192102134228,
+      "learning_rate": 7.692307692307694e-08,
+      "loss": 0.0012,
+      "step": 19925
+    },
+    {
+      "epoch": 21.97136563876652,
+      "grad_norm": 0.00970546342432499,
+      "learning_rate": 5.1282051282051286e-08,
+      "loss": 0.0012,
+      "step": 19950
+    },
+    {
+      "epoch": 21.998898678414097,
+      "grad_norm": 0.0086264219135046,
+      "learning_rate": 2.5641025641025643e-08,
+      "loss": 0.0015,
+      "step": 19975
+    },
+    {
+      "epoch": 22.026431718061673,
+      "grad_norm": 0.008207079023122787,
+      "learning_rate": 0.0,
+      "loss": 0.0012,
+      "step": 20000
+    },
+    {
+      "epoch": 22.026431718061673,
+      "eval_cer": 50.30034505290429,
+      "eval_loss": 0.9635043144226074,
+      "eval_runtime": 839.8575,
+      "eval_samples_per_second": 12.599,
+      "eval_steps_per_second": 3.151,
+      "eval_wer": 94.62517680339462,
+      "step": 20000
+    },
+    {
+      "epoch": 22.026431718061673,
+      "step": 20000,
+      "total_flos": 1.150199251255427e+20,
+      "train_loss": 0.29341357232630255,
+      "train_runtime": 51592.7978,
+      "train_samples_per_second": 12.405,
+      "train_steps_per_second": 0.388
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 20000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 23,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.150199251255427e+20,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}