diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7900 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 20000,
+  "global_step": 130190,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996236270066826e-05,
+      "loss": 3.6825,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992395729318688e-05,
+      "loss": 3.1145,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988555188570551e-05,
+      "loss": 2.9433,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984714647822414e-05,
+      "loss": 2.8288,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9808741070742764e-05,
+      "loss": 2.7201,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977033566326139e-05,
+      "loss": 2.6525,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9731930255780016e-05,
+      "loss": 2.5626,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9693908902373456e-05,
+      "loss": 2.5155,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965550349489208e-05,
+      "loss": 2.4511,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961709808741071e-05,
+      "loss": 2.4267,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.957869267992934e-05,
+      "loss": 2.3938,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954067132652278e-05,
+      "loss": 2.3345,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95022659190414e-05,
+      "loss": 2.2903,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.946386051156003e-05,
+      "loss": 2.2496,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942545510407866e-05,
+      "loss": 2.2153,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.938704969659728e-05,
+      "loss": 2.1876,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.934864428911591e-05,
+      "loss": 2.1494,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.931023888163453e-05,
+      "loss": 2.1008,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927183347415316e-05,
+      "loss": 2.083,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.923342806667179e-05,
+      "loss": 2.0438,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9195022659190414e-05,
+      "loss": 2.0105,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915661725170904e-05,
+      "loss": 1.9819,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9118211844227666e-05,
+      "loss": 1.9548,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9079806436746295e-05,
+      "loss": 1.9269,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9041401029264924e-05,
+      "loss": 1.9,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.900299562178355e-05,
+      "loss": 1.8725,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8964590214302176e-05,
+      "loss": 1.8457,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89261848068208e-05,
+      "loss": 1.8149,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.888777939933943e-05,
+      "loss": 1.798,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.884937399185806e-05,
+      "loss": 1.77,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.881096858437668e-05,
+      "loss": 1.7548,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877256317689531e-05,
+      "loss": 1.7297,
+      "step": 3200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.873415776941393e-05,
+      "loss": 1.6889,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.869575236193256e-05,
+      "loss": 1.6675,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.865734695445119e-05,
+      "loss": 1.6599,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.861894154696981e-05,
+      "loss": 1.6153,
+      "step": 3600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.858053613948844e-05,
+      "loss": 1.5925,
+      "step": 3700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.854213073200707e-05,
+      "loss": 1.5807,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8503725324525694e-05,
+      "loss": 1.5442,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8465319917044324e-05,
+      "loss": 1.5413,
+      "step": 4000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8426914509562946e-05,
+      "loss": 1.5127,
+      "step": 4100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8388509102081576e-05,
+      "loss": 1.4997,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8350103694600205e-05,
+      "loss": 1.4767,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.831169828711883e-05,
+      "loss": 1.4655,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827329287963746e-05,
+      "loss": 1.4414,
+      "step": 4500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.823488747215608e-05,
+      "loss": 1.4253,
+      "step": 4600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.819648206467471e-05,
+      "loss": 1.4103,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815807665719334e-05,
+      "loss": 1.3864,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.811967124971196e-05,
+      "loss": 1.3764,
+      "step": 4900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808126584223059e-05,
+      "loss": 1.342,
+      "step": 5000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.804286043474921e-05,
+      "loss": 1.331,
+      "step": 5100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.800445502726784e-05,
+      "loss": 1.3282,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796604961978647e-05,
+      "loss": 1.3114,
+      "step": 5300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7927644212305094e-05,
+      "loss": 1.2951,
+      "step": 5400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.788923880482372e-05,
+      "loss": 1.279,
+      "step": 5500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7850833397342345e-05,
+      "loss": 1.2618,
+      "step": 5600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7812427989860975e-05,
+      "loss": 1.2417,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7774022582379604e-05,
+      "loss": 1.2182,
+      "step": 5800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7735617174898227e-05,
+      "loss": 1.2171,
+      "step": 5900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7697211767416856e-05,
+      "loss": 1.2039,
+      "step": 6000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.765880635993548e-05,
+      "loss": 1.1928,
+      "step": 6100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762040095245411e-05,
+      "loss": 1.175,
+      "step": 6200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.758199554497274e-05,
+      "loss": 1.1632,
+      "step": 6300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.754359013749136e-05,
+      "loss": 1.1471,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.750518473000999e-05,
+      "loss": 1.1404,
+      "step": 6500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.746677932252862e-05,
+      "loss": 1.1234,
+      "step": 6600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.742837391504724e-05,
+      "loss": 1.1143,
+      "step": 6700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.738996850756587e-05,
+      "loss": 1.0999,
+      "step": 6800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735156310008449e-05,
+      "loss": 1.09,
+      "step": 6900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.731315769260312e-05,
+      "loss": 1.0809,
+      "step": 7000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727475228512175e-05,
+      "loss": 1.0603,
+      "step": 7100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7236346877640374e-05,
+      "loss": 1.0536,
+      "step": 7200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7197941470159e-05,
+      "loss": 1.0413,
+      "step": 7300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7159536062677626e-05,
+      "loss": 1.0351,
+      "step": 7400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7121130655196255e-05,
+      "loss": 1.0306,
+      "step": 7500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7082725247714884e-05,
+      "loss": 1.0009,
+      "step": 7600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.704431984023351e-05,
+      "loss": 1.0067,
+      "step": 7700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7005914432752136e-05,
+      "loss": 0.9934,
+      "step": 7800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.696750902527076e-05,
+      "loss": 0.9772,
+      "step": 7900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.692910361778939e-05,
+      "loss": 0.9657,
+      "step": 8000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.689069821030802e-05,
+      "loss": 0.9626,
+      "step": 8100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685229280282664e-05,
+      "loss": 0.9601,
+      "step": 8200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.681388739534527e-05,
+      "loss": 0.9489,
+      "step": 8300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677548198786389e-05,
+      "loss": 0.9426,
+      "step": 8400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.673707658038252e-05,
+      "loss": 0.9264,
+      "step": 8500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.669867117290115e-05,
+      "loss": 0.9215,
+      "step": 8600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.666026576541977e-05,
+      "loss": 0.9066,
+      "step": 8700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66218603579384e-05,
+      "loss": 0.8992,
+      "step": 8800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6583454950457025e-05,
+      "loss": 0.898,
+      "step": 8900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6545049542975654e-05,
+      "loss": 0.8883,
+      "step": 9000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6506644135494284e-05,
+      "loss": 0.8709,
+      "step": 9100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6468238728012906e-05,
+      "loss": 0.8671,
+      "step": 9200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6429833320531536e-05,
+      "loss": 0.8649,
+      "step": 9300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6391427913050165e-05,
+      "loss": 0.8513,
+      "step": 9400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.635302250556879e-05,
+      "loss": 0.8578,
+      "step": 9500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.631461709808742e-05,
+      "loss": 0.8355,
+      "step": 9600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627621169060604e-05,
+      "loss": 0.8347,
+      "step": 9700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623780628312467e-05,
+      "loss": 0.8265,
+      "step": 9800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.61994008756433e-05,
+      "loss": 0.8204,
+      "step": 9900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616099546816192e-05,
+      "loss": 0.811,
+      "step": 10000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612259006068055e-05,
+      "loss": 0.8196,
+      "step": 10100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.608418465319917e-05,
+      "loss": 0.7999,
+      "step": 10200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60457792457178e-05,
+      "loss": 0.7979,
+      "step": 10300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600737383823643e-05,
+      "loss": 0.7915,
+      "step": 10400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5968968430755054e-05,
+      "loss": 0.7821,
+      "step": 10500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.593056302327368e-05,
+      "loss": 0.7852,
+      "step": 10600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5892157615792305e-05,
+      "loss": 0.7712,
+      "step": 10700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5853752208310935e-05,
+      "loss": 0.7637,
+      "step": 10800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5815346800829564e-05,
+      "loss": 0.7552,
+      "step": 10900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5776941393348187e-05,
+      "loss": 0.7522,
+      "step": 11000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5738535985866816e-05,
+      "loss": 0.7495,
+      "step": 11100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.570013057838544e-05,
+      "loss": 0.7427,
+      "step": 11200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566172517090406e-05,
+      "loss": 0.7357,
+      "step": 11300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562331976342269e-05,
+      "loss": 0.7363,
+      "step": 11400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.558529841001613e-05,
+      "loss": 0.7348,
+      "step": 11500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.554689300253476e-05,
+      "loss": 0.726,
+      "step": 11600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.550848759505338e-05,
+      "loss": 0.7096,
+      "step": 11700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547008218757201e-05,
+      "loss": 0.7048,
+      "step": 11800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5431676780090634e-05,
+      "loss": 0.7129,
+      "step": 11900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5393271372609263e-05,
+      "loss": 0.7074,
+      "step": 12000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.535486596512789e-05,
+      "loss": 0.6931,
+      "step": 12100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5316460557646515e-05,
+      "loss": 0.6922,
+      "step": 12200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5278055150165145e-05,
+      "loss": 0.6819,
+      "step": 12300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5239649742683774e-05,
+      "loss": 0.683,
+      "step": 12400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5201244335202397e-05,
+      "loss": 0.6804,
+      "step": 12500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5162838927721026e-05,
+      "loss": 0.6745,
+      "step": 12600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512443352023965e-05,
+      "loss": 0.6662,
+      "step": 12700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.508602811275828e-05,
+      "loss": 0.6619,
+      "step": 12800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.504762270527691e-05,
+      "loss": 0.6622,
+      "step": 12900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.500921729779553e-05,
+      "loss": 0.6539,
+      "step": 13000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497119594438897e-05,
+      "loss": 0.6572,
+      "step": 13100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.49327905369076e-05,
+      "loss": 0.6299,
+      "step": 13200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.489438512942623e-05,
+      "loss": 0.6188,
+      "step": 13300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.485597972194485e-05,
+      "loss": 0.6245,
+      "step": 13400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481757431446348e-05,
+      "loss": 0.6171,
+      "step": 13500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.47791689069821e-05,
+      "loss": 0.6094,
+      "step": 13600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.474076349950073e-05,
+      "loss": 0.6123,
+      "step": 13700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.470235809201936e-05,
+      "loss": 0.6086,
+      "step": 13800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4663952684537984e-05,
+      "loss": 0.6067,
+      "step": 13900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462554727705661e-05,
+      "loss": 0.6021,
+      "step": 14000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4587141869575236e-05,
+      "loss": 0.5989,
+      "step": 14100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4548736462093865e-05,
+      "loss": 0.5908,
+      "step": 14200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4510331054612495e-05,
+      "loss": 0.5911,
+      "step": 14300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447192564713112e-05,
+      "loss": 0.5861,
+      "step": 14400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4433520239649746e-05,
+      "loss": 0.587,
+      "step": 14500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.439511483216837e-05,
+      "loss": 0.5802,
+      "step": 14600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4356709424687e-05,
+      "loss": 0.5712,
+      "step": 14700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431830401720563e-05,
+      "loss": 0.5733,
+      "step": 14800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.427989860972425e-05,
+      "loss": 0.5691,
+      "step": 14900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.424149320224288e-05,
+      "loss": 0.5643,
+      "step": 15000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42030877947615e-05,
+      "loss": 0.5688,
+      "step": 15100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416468238728013e-05,
+      "loss": 0.5677,
+      "step": 15200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412627697979876e-05,
+      "loss": 0.5588,
+      "step": 15300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.408787157231738e-05,
+      "loss": 0.5639,
+      "step": 15400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.404946616483601e-05,
+      "loss": 0.5561,
+      "step": 15500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.401106075735464e-05,
+      "loss": 0.5519,
+      "step": 15600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3972655349873264e-05,
+      "loss": 0.5522,
+      "step": 15700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3934249942391894e-05,
+      "loss": 0.5476,
+      "step": 15800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3895844534910516e-05,
+      "loss": 0.541,
+      "step": 15900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3857439127429146e-05,
+      "loss": 0.5464,
+      "step": 16000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3819033719947775e-05,
+      "loss": 0.5402,
+      "step": 16100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.37806283124664e-05,
+      "loss": 0.5364,
+      "step": 16200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.374222290498503e-05,
+      "loss": 0.5385,
+      "step": 16300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.370381749750365e-05,
+      "loss": 0.5308,
+      "step": 16400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366541209002228e-05,
+      "loss": 0.5301,
+      "step": 16500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362700668254091e-05,
+      "loss": 0.5298,
+      "step": 16600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358860127505953e-05,
+      "loss": 0.5244,
+      "step": 16700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.355019586757816e-05,
+      "loss": 0.5299,
+      "step": 16800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351179046009678e-05,
+      "loss": 0.5204,
+      "step": 16900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347338505261541e-05,
+      "loss": 0.5166,
+      "step": 17000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.343497964513404e-05,
+      "loss": 0.5199,
+      "step": 17100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3396574237652664e-05,
+      "loss": 0.5187,
+      "step": 17200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.335816883017129e-05,
+      "loss": 0.5118,
+      "step": 17300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3319763422689915e-05,
+      "loss": 0.5091,
+      "step": 17400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3281358015208545e-05,
+      "loss": 0.5135,
+      "step": 17500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3242952607727174e-05,
+      "loss": 0.5063,
+      "step": 17600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32045472002458e-05,
+      "loss": 0.4981,
+      "step": 17700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3166141792764426e-05,
+      "loss": 0.5012,
+      "step": 17800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.312773638528305e-05,
+      "loss": 0.5027,
+      "step": 17900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.308933097780168e-05,
+      "loss": 0.4948,
+      "step": 18000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.305092557032031e-05,
+      "loss": 0.4962,
+      "step": 18100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301252016283893e-05,
+      "loss": 0.4928,
+      "step": 18200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297411475535756e-05,
+      "loss": 0.4989,
+      "step": 18300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293570934787619e-05,
+      "loss": 0.4881,
+      "step": 18400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.289730394039481e-05,
+      "loss": 0.4881,
+      "step": 18500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.285889853291344e-05,
+      "loss": 0.4855,
+      "step": 18600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282049312543206e-05,
+      "loss": 0.4818,
+      "step": 18700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.278208771795069e-05,
+      "loss": 0.4829,
+      "step": 18800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2744066364544125e-05,
+      "loss": 0.481,
+      "step": 18900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2705660957062755e-05,
+      "loss": 0.4745,
+      "step": 19000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2667255549581384e-05,
+      "loss": 0.4758,
+      "step": 19100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.262885014210001e-05,
+      "loss": 0.479,
+      "step": 19200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2590444734618636e-05,
+      "loss": 0.4788,
+      "step": 19300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.255203932713726e-05,
+      "loss": 0.4765,
+      "step": 19400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251363391965589e-05,
+      "loss": 0.4672,
+      "step": 19500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247522851217452e-05,
+      "loss": 0.467,
+      "step": 19600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.243682310469314e-05,
+      "loss": 0.4642,
+      "step": 19700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.239841769721177e-05,
+      "loss": 0.4679,
+      "step": 19800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236001228973039e-05,
+      "loss": 0.4656,
+      "step": 19900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232160688224902e-05,
+      "loss": 0.4548,
+      "step": 20000
+    },
+    {
+      "epoch": 1.54,
+      "eval_cer": 2.8856,
+      "eval_gen_len": 12.78068,
+      "eval_loss": 0.4116576611995697,
+      "eval_runtime": 1412.1401,
+      "eval_samples_per_second": 70.815,
+      "eval_steps_per_second": 0.277,
+      "eval_wer": 13.6181,
+      "step": 20000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.228320147476765e-05,
+      "loss": 0.4658,
+      "step": 20100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.224479606728627e-05,
+      "loss": 0.4507,
+      "step": 20200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22063906598049e-05,
+      "loss": 0.4552,
+      "step": 20300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2167985252323525e-05,
+      "loss": 0.4526,
+      "step": 20400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2129579844842154e-05,
+      "loss": 0.4458,
+      "step": 20500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.209117443736078e-05,
+      "loss": 0.4548,
+      "step": 20600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2052769029879406e-05,
+      "loss": 0.4467,
+      "step": 20700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2014363622398035e-05,
+      "loss": 0.4481,
+      "step": 20800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197595821491666e-05,
+      "loss": 0.4477,
+      "step": 20900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.193755280743529e-05,
+      "loss": 0.4446,
+      "step": 21000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1899147399953916e-05,
+      "loss": 0.4435,
+      "step": 21100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186074199247254e-05,
+      "loss": 0.4381,
+      "step": 21200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182233658499117e-05,
+      "loss": 0.4432,
+      "step": 21300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.178393117750979e-05,
+      "loss": 0.4365,
+      "step": 21400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.174552577002842e-05,
+      "loss": 0.4383,
+      "step": 21500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.170712036254705e-05,
+      "loss": 0.4368,
+      "step": 21600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166871495506567e-05,
+      "loss": 0.4351,
+      "step": 21700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.16303095475843e-05,
+      "loss": 0.4306,
+      "step": 21800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.159190414010293e-05,
+      "loss": 0.4315,
+      "step": 21900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.155349873262155e-05,
+      "loss": 0.4335,
+      "step": 22000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151509332514018e-05,
+      "loss": 0.4258,
+      "step": 22100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1476687917658805e-05,
+      "loss": 0.4326,
+      "step": 22200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1438282510177434e-05,
+      "loss": 0.4298,
+      "step": 22300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1399877102696064e-05,
+      "loss": 0.4208,
+      "step": 22400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1361471695214686e-05,
+      "loss": 0.4244,
+      "step": 22500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1323066287733316e-05,
+      "loss": 0.4248,
+      "step": 22600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.128466088025194e-05,
+      "loss": 0.4285,
+      "step": 22700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.124625547277057e-05,
+      "loss": 0.421,
+      "step": 22800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12078500652892e-05,
+      "loss": 0.4225,
+      "step": 22900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116944465780782e-05,
+      "loss": 0.416,
+      "step": 23000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.113103925032645e-05,
+      "loss": 0.4127,
+      "step": 23100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.109263384284507e-05,
+      "loss": 0.4173,
+      "step": 23200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.10542284353637e-05,
+      "loss": 0.4161,
+      "step": 23300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101582302788233e-05,
+      "loss": 0.414,
+      "step": 23400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097741762040095e-05,
+      "loss": 0.4106,
+      "step": 23500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.093901221291958e-05,
+      "loss": 0.4102,
+      "step": 23600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0900606805438204e-05,
+      "loss": 0.408,
+      "step": 23700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0862201397956834e-05,
+      "loss": 0.411,
+      "step": 23800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0824180044550274e-05,
+      "loss": 0.4103,
+      "step": 23900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.07857746370689e-05,
+      "loss": 0.4085,
+      "step": 24000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0747369229587526e-05,
+      "loss": 0.4087,
+      "step": 24100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0708963822106155e-05,
+      "loss": 0.404,
+      "step": 24200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0670558414624784e-05,
+      "loss": 0.4018,
+      "step": 24300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.063215300714341e-05,
+      "loss": 0.4052,
+      "step": 24400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0593747599662036e-05,
+      "loss": 0.4056,
+      "step": 24500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.055534219218066e-05,
+      "loss": 0.4054,
+      "step": 24600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051693678469929e-05,
+      "loss": 0.397,
+      "step": 24700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.047853137721792e-05,
+      "loss": 0.3996,
+      "step": 24800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.044012596973654e-05,
+      "loss": 0.3976,
+      "step": 24900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.040172056225517e-05,
+      "loss": 0.3914,
+      "step": 25000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03633151547738e-05,
+      "loss": 0.3955,
+      "step": 25100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032490974729242e-05,
+      "loss": 0.3959,
+      "step": 25200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.028688839388586e-05,
+      "loss": 0.3927,
+      "step": 25300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.02488670404793e-05,
+      "loss": 0.3928,
+      "step": 25400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0210461632997924e-05,
+      "loss": 0.392,
+      "step": 25500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017205622551655e-05,
+      "loss": 0.3895,
+      "step": 25600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0133650818035176e-05,
+      "loss": 0.3918,
+      "step": 25700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0095245410553805e-05,
+      "loss": 0.3855,
+      "step": 25800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0056840003072434e-05,
+      "loss": 0.3862,
+      "step": 25900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001843459559106e-05,
+      "loss": 0.388,
+      "step": 26000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9980029188109686e-05,
+      "loss": 0.3697,
+      "step": 26100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9941623780628316e-05,
+      "loss": 0.3568,
+      "step": 26200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.990321837314694e-05,
+      "loss": 0.3631,
+      "step": 26300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986481296566557e-05,
+      "loss": 0.3552,
+      "step": 26400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982640755818419e-05,
+      "loss": 0.355,
+      "step": 26500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.978800215070282e-05,
+      "loss": 0.3577,
+      "step": 26600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.974959674322145e-05,
+      "loss": 0.3615,
+      "step": 26700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971119133574007e-05,
+      "loss": 0.3557,
+      "step": 26800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.96727859282587e-05,
+      "loss": 0.3609,
+      "step": 26900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.963438052077732e-05,
+      "loss": 0.3593,
+      "step": 27000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.959597511329595e-05,
+      "loss": 0.3567,
+      "step": 27100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.955756970581458e-05,
+      "loss": 0.3559,
+      "step": 27200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9519164298333204e-05,
+      "loss": 0.3558,
+      "step": 27300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9480758890851834e-05,
+      "loss": 0.3567,
+      "step": 27400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9442353483370456e-05,
+      "loss": 0.3565,
+      "step": 27500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9403948075889085e-05,
+      "loss": 0.3529,
+      "step": 27600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9365542668407715e-05,
+      "loss": 0.3563,
+      "step": 27700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932713726092634e-05,
+      "loss": 0.351,
+      "step": 27800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9288731853444967e-05,
+      "loss": 0.3538,
+      "step": 27900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925032644596359e-05,
+      "loss": 0.3546,
+      "step": 28000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921192103848222e-05,
+      "loss": 0.3519,
+      "step": 28100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917351563100085e-05,
+      "loss": 0.3502,
+      "step": 28200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.913511022351947e-05,
+      "loss": 0.3476,
+      "step": 28300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.90967048160381e-05,
+      "loss": 0.3565,
+      "step": 28400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.905829940855672e-05,
+      "loss": 0.351,
+      "step": 28500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901989400107535e-05,
+      "loss": 0.3453,
+      "step": 28600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.898148859359398e-05,
+      "loss": 0.3459,
+      "step": 28700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89430831861126e-05,
+      "loss": 0.3505,
+      "step": 28800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.890467777863123e-05,
+      "loss": 0.3454,
+      "step": 28900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886627237114986e-05,
+      "loss": 0.3488,
+      "step": 29000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8827866963668485e-05,
+      "loss": 0.3466,
+      "step": 29100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8789461556187114e-05,
+      "loss": 0.3445,
+      "step": 29200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8751056148705736e-05,
+      "loss": 0.3431,
+      "step": 29300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8712650741224366e-05,
+      "loss": 0.3418,
+      "step": 29400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8674245333742995e-05,
+      "loss": 0.3388,
+      "step": 29500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.863583992626162e-05,
+      "loss": 0.343,
+      "step": 29600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.859743451878025e-05,
+      "loss": 0.3417,
+      "step": 29700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.855902911129887e-05,
+      "loss": 0.339,
+      "step": 29800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85206237038175e-05,
+      "loss": 0.3412,
+      "step": 29900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.848221829633613e-05,
+      "loss": 0.3405,
+      "step": 30000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.844381288885475e-05,
+      "loss": 0.3434,
+      "step": 30100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.840540748137338e-05,
+      "loss": 0.3378,
+      "step": 30200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8367002073892e-05,
+      "loss": 0.3366,
+      "step": 30300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832859666641063e-05,
+      "loss": 0.3401,
+      "step": 30400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.829019125892926e-05,
+      "loss": 0.3405,
+      "step": 30500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8251785851447884e-05,
+      "loss": 0.3395,
+      "step": 30600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821338044396651e-05,
+      "loss": 0.3377,
+      "step": 30700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8174975036485136e-05,
+      "loss": 0.332,
+      "step": 30800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8136569629003765e-05,
+      "loss": 0.3357,
+      "step": 30900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8098164221522394e-05,
+      "loss": 0.3307,
+      "step": 31000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.805975881404102e-05,
+      "loss": 0.3377,
+      "step": 31100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8021353406559646e-05,
+      "loss": 0.3344,
+      "step": 31200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.7982947999078275e-05,
+      "loss": 0.3287,
+      "step": 31300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.79445425915969e-05,
+      "loss": 0.3327,
+      "step": 31400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.790613718411553e-05,
+      "loss": 0.3324,
+      "step": 31500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786773177663415e-05,
+      "loss": 0.3296,
+      "step": 31600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.782932636915278e-05,
+      "loss": 0.3333,
+      "step": 31700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.779092096167141e-05,
+      "loss": 0.3361,
+      "step": 31800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775251555419003e-05,
+      "loss": 0.3273,
+      "step": 31900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771411014670866e-05,
+      "loss": 0.328,
+      "step": 32000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767570473922728e-05,
+      "loss": 0.3281,
+      "step": 32100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.763729933174591e-05,
+      "loss": 0.3263,
+      "step": 32200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.759889392426454e-05,
+      "loss": 0.3249,
+      "step": 32300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7560488516783164e-05,
+      "loss": 0.3269,
+      "step": 32400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7522083109301793e-05,
+      "loss": 0.3238,
+      "step": 32500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7483677701820416e-05,
+      "loss": 0.3251,
+      "step": 32600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744565634841386e-05,
+      "loss": 0.3271,
+      "step": 32700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7407250940932485e-05,
+      "loss": 0.3229,
+      "step": 32800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7368845533451115e-05,
+      "loss": 0.321,
+      "step": 32900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.733044012596974e-05,
+      "loss": 0.3239,
+      "step": 33000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.729203471848837e-05,
+      "loss": 0.322,
+      "step": 33100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7253629311006996e-05,
+      "loss": 0.3218,
+      "step": 33200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721522390352562e-05,
+      "loss": 0.3235,
+      "step": 33300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717681849604425e-05,
+      "loss": 0.3244,
+      "step": 33400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.713841308856287e-05,
+      "loss": 0.3223,
+      "step": 33500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71000076810815e-05,
+      "loss": 0.3189,
+      "step": 33600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706198632767493e-05,
+      "loss": 0.3187,
+      "step": 33700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702358092019356e-05,
+      "loss": 0.3199,
+      "step": 33800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698517551271219e-05,
+      "loss": 0.3241,
+      "step": 33900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6946770105230814e-05,
+      "loss": 0.3172,
+      "step": 34000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6908364697749444e-05,
+      "loss": 0.3201,
+      "step": 34100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6869959290268066e-05,
+      "loss": 0.3183,
+      "step": 34200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6831553882786695e-05,
+      "loss": 0.3142,
+      "step": 34300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6793148475305325e-05,
+      "loss": 0.317,
+      "step": 34400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.675474306782395e-05,
+      "loss": 0.3158,
+      "step": 34500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671633766034258e-05,
+      "loss": 0.3176,
+      "step": 34600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.66779322528612e-05,
+      "loss": 0.3153,
+      "step": 34700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.663952684537983e-05,
+      "loss": 0.3142,
+      "step": 34800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.660112143789846e-05,
+      "loss": 0.3118,
+      "step": 34900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.656271603041708e-05,
+      "loss": 0.3121,
+      "step": 35000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652431062293571e-05,
+      "loss": 0.3113,
+      "step": 35100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.648590521545434e-05,
+      "loss": 0.314,
+      "step": 35200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.644749980797296e-05,
+      "loss": 0.3156,
+      "step": 35300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.640909440049159e-05,
+      "loss": 0.3135,
+      "step": 35400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6370688993010213e-05,
+      "loss": 0.3151,
+      "step": 35500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.633228358552884e-05,
+      "loss": 0.3124,
+      "step": 35600
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.629387817804747e-05,
+      "loss": 0.3092,
+      "step": 35700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6255472770566095e-05,
+      "loss": 0.3089,
+      "step": 35800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6217067363084724e-05,
+      "loss": 0.3149,
+      "step": 35900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6178661955603347e-05,
+      "loss": 0.3114,
+      "step": 36000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6140256548121976e-05,
+      "loss": 0.3077,
+      "step": 36100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6101851140640605e-05,
+      "loss": 0.3069,
+      "step": 36200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606344573315923e-05,
+      "loss": 0.3097,
+      "step": 36300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602504032567786e-05,
+      "loss": 0.3121,
+      "step": 36400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.598663491819648e-05,
+      "loss": 0.3119,
+      "step": 36500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.594822951071511e-05,
+      "loss": 0.3086,
+      "step": 36600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.590982410323374e-05,
+      "loss": 0.3071,
+      "step": 36700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587141869575236e-05,
+      "loss": 0.3057,
+      "step": 36800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.583301328827099e-05,
+      "loss": 0.3077,
+      "step": 36900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.579460788078961e-05,
+      "loss": 0.3066,
+      "step": 37000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.575620247330824e-05,
+      "loss": 0.3059,
+      "step": 37100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571779706582687e-05,
+      "loss": 0.305,
+      "step": 37200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5679391658345494e-05,
+      "loss": 0.3021,
+      "step": 37300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.564098625086412e-05,
+      "loss": 0.3011,
+      "step": 37400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5602580843382746e-05,
+      "loss": 0.3011,
+      "step": 37500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5564175435901375e-05,
+      "loss": 0.3013,
+      "step": 37600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5525770028420004e-05,
+      "loss": 0.3016,
+      "step": 37700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.548736462093863e-05,
+      "loss": 0.3066,
+      "step": 37800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5448959213457256e-05,
+      "loss": 0.3033,
+      "step": 37900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5410553805975886e-05,
+      "loss": 0.3029,
+      "step": 38000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537214839849451e-05,
+      "loss": 0.304,
+      "step": 38100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.533374299101314e-05,
+      "loss": 0.3008,
+      "step": 38200
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.529533758353176e-05,
+      "loss": 0.3003,
+      "step": 38300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.525693217605039e-05,
+      "loss": 0.2956,
+      "step": 38400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521852676856902e-05,
+      "loss": 0.2998,
+      "step": 38500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.518012136108764e-05,
+      "loss": 0.2996,
+      "step": 38600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.514171595360627e-05,
+      "loss": 0.3001,
+      "step": 38700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510331054612489e-05,
+      "loss": 0.295,
+      "step": 38800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506490513864352e-05,
+      "loss": 0.2972,
+      "step": 38900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502649973116215e-05,
+      "loss": 0.2969,
+      "step": 39000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4988094323680774e-05,
+      "loss": 0.2868,
+      "step": 39100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4949688916199404e-05,
+      "loss": 0.2754,
+      "step": 39200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4911283508718026e-05,
+      "loss": 0.2746,
+      "step": 39300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4872878101236655e-05,
+      "loss": 0.2756,
+      "step": 39400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4834472693755285e-05,
+      "loss": 0.2751,
+      "step": 39500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.479606728627391e-05,
+      "loss": 0.2753,
+      "step": 39600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.475766187879254e-05,
+      "loss": 0.273,
+      "step": 39700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471925647131116e-05,
+      "loss": 0.2738,
+      "step": 39800
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.468085106382979e-05,
+      "loss": 0.2752,
+      "step": 39900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.464244565634842e-05,
+      "loss": 0.2723,
+      "step": 40000
+    },
+    {
+      "epoch": 3.07,
+      "eval_cer": 2.1004,
+      "eval_gen_len": 12.68084,
+      "eval_loss": 0.27356424927711487,
+      "eval_runtime": 1382.5597,
+      "eval_samples_per_second": 72.33,
+      "eval_steps_per_second": 0.283,
+      "eval_wer": 10.5883,
+      "step": 40000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.460404024886704e-05,
+      "loss": 0.2739,
+      "step": 40100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456563484138567e-05,
+      "loss": 0.2741,
+      "step": 40200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45272294339043e-05,
+      "loss": 0.2742,
+      "step": 40300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.448882402642292e-05,
+      "loss": 0.275,
+      "step": 40400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.445041861894155e-05,
+      "loss": 0.2706,
+      "step": 40500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441239726553499e-05,
+      "loss": 0.269,
+      "step": 40600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4373991858053614e-05,
+      "loss": 0.2714,
+      "step": 40700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.433558645057224e-05,
+      "loss": 0.2747,
+      "step": 40800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.429718104309087e-05,
+      "loss": 0.2712,
+      "step": 40900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4258775635609495e-05,
+      "loss": 0.2713,
+      "step": 41000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4220370228128124e-05,
+      "loss": 0.2738,
+      "step": 41100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4181964820646753e-05,
+      "loss": 0.2713,
+      "step": 41200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4143559413165376e-05,
+      "loss": 0.2736,
+      "step": 41300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4105154005684005e-05,
+      "loss": 0.2724,
+      "step": 41400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406674859820263e-05,
+      "loss": 0.2695,
+      "step": 41500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402834319072126e-05,
+      "loss": 0.2713,
+      "step": 41600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3989937783239886e-05,
+      "loss": 0.2737,
+      "step": 41700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395153237575851e-05,
+      "loss": 0.2706,
+      "step": 41800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391312696827714e-05,
+      "loss": 0.2738,
+      "step": 41900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.387472156079576e-05,
+      "loss": 0.2718,
+      "step": 42000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.383631615331439e-05,
+      "loss": 0.2705,
+      "step": 42100
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.379791074583302e-05,
+      "loss": 0.2718,
+      "step": 42200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.375950533835164e-05,
+      "loss": 0.2723,
+      "step": 42300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.372109993087027e-05,
+      "loss": 0.2692,
+      "step": 42400
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3682694523388894e-05,
+      "loss": 0.2662,
+      "step": 42500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.364428911590752e-05,
+      "loss": 0.2709,
+      "step": 42600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.360588370842615e-05,
+      "loss": 0.2732,
+      "step": 42700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3567478300944775e-05,
+      "loss": 0.2695,
+      "step": 42800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3529072893463404e-05,
+      "loss": 0.2714,
+      "step": 42900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.349066748598203e-05,
+      "loss": 0.268,
+      "step": 43000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3452262078500656e-05,
+      "loss": 0.2701,
+      "step": 43100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3413856671019286e-05,
+      "loss": 0.2661,
+      "step": 43200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337545126353791e-05,
+      "loss": 0.2704,
+      "step": 43300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.333704585605654e-05,
+      "loss": 0.2683,
+      "step": 43400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.329864044857517e-05,
+      "loss": 0.2679,
+      "step": 43500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326023504109379e-05,
+      "loss": 0.2654,
+      "step": 43600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322182963361242e-05,
+      "loss": 0.2713,
+      "step": 43700
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.318342422613104e-05,
+      "loss": 0.2694,
+      "step": 43800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.314501881864967e-05,
+      "loss": 0.2676,
+      "step": 43900
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.31066134111683e-05,
+      "loss": 0.2653,
+      "step": 44000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3068208003686916e-05,
+      "loss": 0.2691,
+      "step": 44100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3029802596205545e-05,
+      "loss": 0.2668,
+      "step": 44200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2991397188724174e-05,
+      "loss": 0.2686,
+      "step": 44300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.29529917812428e-05,
+      "loss": 0.265,
+      "step": 44400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2914586373761426e-05,
+      "loss": 0.2688,
+      "step": 44500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2876565020354866e-05,
+      "loss": 0.2648,
+      "step": 44600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2838159612873496e-05,
+      "loss": 0.2653,
+      "step": 44700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.279975420539212e-05,
+      "loss": 0.2679,
+      "step": 44800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276134879791075e-05,
+      "loss": 0.2662,
+      "step": 44900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272294339042937e-05,
+      "loss": 0.2645,
+      "step": 45000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2684537982948e-05,
+      "loss": 0.2706,
+      "step": 45100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264613257546663e-05,
+      "loss": 0.2624,
+      "step": 45200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.260772716798525e-05,
+      "loss": 0.2626,
+      "step": 45300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256932176050388e-05,
+      "loss": 0.2657,
+      "step": 45400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25309163530225e-05,
+      "loss": 0.2628,
+      "step": 45500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.249251094554113e-05,
+      "loss": 0.2641,
+      "step": 45600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.245410553805976e-05,
+      "loss": 0.2625,
+      "step": 45700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2415700130578384e-05,
+      "loss": 0.2614,
+      "step": 45800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2377294723097014e-05,
+      "loss": 0.2614,
+      "step": 45900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2338889315615636e-05,
+      "loss": 0.2611,
+      "step": 46000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2300483908134266e-05,
+      "loss": 0.2663,
+      "step": 46100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2262078500652895e-05,
+      "loss": 0.2626,
+      "step": 46200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222367309317152e-05,
+      "loss": 0.258,
+      "step": 46300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.218526768569015e-05,
+      "loss": 0.2612,
+      "step": 46400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.214686227820877e-05,
+      "loss": 0.263,
+      "step": 46500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2108840924802216e-05,
+      "loss": 0.2656,
+      "step": 46600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207043551732084e-05,
+      "loss": 0.2608,
+      "step": 46700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.203203010983947e-05,
+      "loss": 0.2608,
+      "step": 46800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.199362470235809e-05,
+      "loss": 0.2615,
+      "step": 46900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.195598740302635e-05,
+      "loss": 0.258,
+      "step": 47000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.191758199554498e-05,
+      "loss": 0.2579,
+      "step": 47100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.18791765880636e-05,
+      "loss": 0.2627,
+      "step": 47200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.184077118058223e-05,
+      "loss": 0.2629,
+      "step": 47300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180236577310086e-05,
+      "loss": 0.2602,
+      "step": 47400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176396036561948e-05,
+      "loss": 0.259,
+      "step": 47500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.172555495813811e-05,
+      "loss": 0.2567,
+      "step": 47600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.168714955065673e-05,
+      "loss": 0.2577,
+      "step": 47700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.164874414317536e-05,
+      "loss": 0.2585,
+      "step": 47800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161033873569399e-05,
+      "loss": 0.2608,
+      "step": 47900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1571933328212614e-05,
+      "loss": 0.2537,
+      "step": 48000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1533527920731244e-05,
+      "loss": 0.2596,
+      "step": 48100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1495122513249866e-05,
+      "loss": 0.2556,
+      "step": 48200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1456717105768496e-05,
+      "loss": 0.2583,
+      "step": 48300
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1418311698287125e-05,
+      "loss": 0.2606,
+      "step": 48400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.137990629080575e-05,
+      "loss": 0.2587,
+      "step": 48500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.134150088332438e-05,
+      "loss": 0.256,
+      "step": 48600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1303095475843006e-05,
+      "loss": 0.2584,
+      "step": 48700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126469006836163e-05,
+      "loss": 0.258,
+      "step": 48800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.122628466088026e-05,
+      "loss": 0.2572,
+      "step": 48900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.118787925339888e-05,
+      "loss": 0.2566,
+      "step": 49000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.11494738459175e-05,
+      "loss": 0.2601,
+      "step": 49100
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111106843843613e-05,
+      "loss": 0.2567,
+      "step": 49200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1072663030954755e-05,
+      "loss": 0.2546,
+      "step": 49300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1034257623473384e-05,
+      "loss": 0.2568,
+      "step": 49400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.0995852215992014e-05,
+      "loss": 0.2527,
+      "step": 49500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0957446808510636e-05,
+      "loss": 0.2581,
+      "step": 49600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0919041401029265e-05,
+      "loss": 0.2553,
+      "step": 49700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.088063599354789e-05,
+      "loss": 0.2573,
+      "step": 49800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.084223058606652e-05,
+      "loss": 0.2568,
+      "step": 49900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080382517858515e-05,
+      "loss": 0.2542,
+      "step": 50000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076541977110377e-05,
+      "loss": 0.2546,
+      "step": 50100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07270143636224e-05,
+      "loss": 0.2556,
+      "step": 50200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.068860895614102e-05,
+      "loss": 0.2518,
+      "step": 50300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.065020354865965e-05,
+      "loss": 0.2539,
+      "step": 50400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061179814117828e-05,
+      "loss": 0.2571,
+      "step": 50500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.05733927336969e-05,
+      "loss": 0.2545,
+      "step": 50600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.053498732621553e-05,
+      "loss": 0.2553,
+      "step": 50700
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0496581918734158e-05,
+      "loss": 0.255,
+      "step": 50800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0458176511252783e-05,
+      "loss": 0.255,
+      "step": 50900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041977110377141e-05,
+      "loss": 0.2548,
+      "step": 51000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038136569629004e-05,
+      "loss": 0.2491,
+      "step": 51100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0342960288808665e-05,
+      "loss": 0.2524,
+      "step": 51200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.030455488132729e-05,
+      "loss": 0.2515,
+      "step": 51300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0266149473845917e-05,
+      "loss": 0.2516,
+      "step": 51400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0227744066364542e-05,
+      "loss": 0.2505,
+      "step": 51500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0189338658883172e-05,
+      "loss": 0.2512,
+      "step": 51600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0150933251401798e-05,
+      "loss": 0.2533,
+      "step": 51700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0112527843920424e-05,
+      "loss": 0.2525,
+      "step": 51800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007412243643905e-05,
+      "loss": 0.2478,
+      "step": 51900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.003571702895768e-05,
+      "loss": 0.2489,
+      "step": 52000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9997311621476305e-05,
+      "loss": 0.2451,
+      "step": 52100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.995890621399493e-05,
+      "loss": 0.2309,
+      "step": 52200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9920500806513557e-05,
+      "loss": 0.2279,
+      "step": 52300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9882095399032183e-05,
+      "loss": 0.2284,
+      "step": 52400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9844074045625626e-05,
+      "loss": 0.2285,
+      "step": 52500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9805668638144252e-05,
+      "loss": 0.2303,
+      "step": 52600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9767263230662878e-05,
+      "loss": 0.2312,
+      "step": 52700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9728857823181504e-05,
+      "loss": 0.2312,
+      "step": 52800
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9690452415700133e-05,
+      "loss": 0.2327,
+      "step": 52900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965204700821876e-05,
+      "loss": 0.2302,
+      "step": 53000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9613641600737385e-05,
+      "loss": 0.2285,
+      "step": 53100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.957523619325601e-05,
+      "loss": 0.2286,
+      "step": 53200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9536830785774637e-05,
+      "loss": 0.2298,
+      "step": 53300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9498425378293266e-05,
+      "loss": 0.2303,
+      "step": 53400
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9460019970811892e-05,
+      "loss": 0.231,
+      "step": 53500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9421614563330518e-05,
+      "loss": 0.2331,
+      "step": 53600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9383209155849144e-05,
+      "loss": 0.235,
+      "step": 53700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.934480374836777e-05,
+      "loss": 0.2319,
+      "step": 53800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9306782394961214e-05,
+      "loss": 0.2305,
+      "step": 53900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.926837698747984e-05,
+      "loss": 0.2301,
+      "step": 54000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9229971579998466e-05,
+      "loss": 0.2286,
+      "step": 54100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.919156617251709e-05,
+      "loss": 0.2288,
+      "step": 54200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.915316076503572e-05,
+      "loss": 0.2332,
+      "step": 54300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9114755357554347e-05,
+      "loss": 0.229,
+      "step": 54400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9076349950072973e-05,
+      "loss": 0.2256,
+      "step": 54500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.90379445425916e-05,
+      "loss": 0.2336,
+      "step": 54600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.8999539135110228e-05,
+      "loss": 0.2307,
+      "step": 54700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8961133727628854e-05,
+      "loss": 0.2297,
+      "step": 54800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892272832014748e-05,
+      "loss": 0.2297,
+      "step": 54900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8884322912666106e-05,
+      "loss": 0.2323,
+      "step": 55000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.884591750518473e-05,
+      "loss": 0.2327,
+      "step": 55100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.880751209770336e-05,
+      "loss": 0.2301,
+      "step": 55200
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8769106690221987e-05,
+      "loss": 0.2294,
+      "step": 55300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8730701282740613e-05,
+      "loss": 0.227,
+      "step": 55400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.869229587525924e-05,
+      "loss": 0.2303,
+      "step": 55500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8653890467777865e-05,
+      "loss": 0.2296,
+      "step": 55600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8615485060296494e-05,
+      "loss": 0.2308,
+      "step": 55700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.857707965281512e-05,
+      "loss": 0.23,
+      "step": 55800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8538674245333746e-05,
+      "loss": 0.226,
+      "step": 55900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8500268837852372e-05,
+      "loss": 0.2311,
+      "step": 56000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8461863430371e-05,
+      "loss": 0.231,
+      "step": 56100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8423458022889627e-05,
+      "loss": 0.2305,
+      "step": 56200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8385052615408253e-05,
+      "loss": 0.2314,
+      "step": 56300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.834664720792688e-05,
+      "loss": 0.2299,
+      "step": 56400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8308241800445505e-05,
+      "loss": 0.2303,
+      "step": 56500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8269836392964134e-05,
+      "loss": 0.2278,
+      "step": 56600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.823143098548276e-05,
+      "loss": 0.226,
+      "step": 56700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8193025578001386e-05,
+      "loss": 0.2302,
+      "step": 56800
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8154620170520012e-05,
+      "loss": 0.2285,
+      "step": 56900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811621476303864e-05,
+      "loss": 0.2293,
+      "step": 57000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8077809355557267e-05,
+      "loss": 0.2243,
+      "step": 57100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8039403948075893e-05,
+      "loss": 0.2294,
+      "step": 57200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800099854059452e-05,
+      "loss": 0.2269,
+      "step": 57300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7962593133113145e-05,
+      "loss": 0.2252,
+      "step": 57400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7924187725631774e-05,
+      "loss": 0.2278,
+      "step": 57500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.78857823181504e-05,
+      "loss": 0.2262,
+      "step": 57600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.784737691066902e-05,
+      "loss": 0.2293,
+      "step": 57700
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.780897150318765e-05,
+      "loss": 0.229,
+      "step": 57800
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7770566095706275e-05,
+      "loss": 0.2265,
+      "step": 57900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.77321606882249e-05,
+      "loss": 0.2292,
+      "step": 58000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7693755280743527e-05,
+      "loss": 0.225,
+      "step": 58100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7655349873262153e-05,
+      "loss": 0.2257,
+      "step": 58200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7616944465780782e-05,
+      "loss": 0.2296,
+      "step": 58300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7578539058299408e-05,
+      "loss": 0.2241,
+      "step": 58400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7540133650818034e-05,
+      "loss": 0.2241,
+      "step": 58500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750172824333666e-05,
+      "loss": 0.2273,
+      "step": 58600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746332283585529e-05,
+      "loss": 0.2236,
+      "step": 58700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7424917428373915e-05,
+      "loss": 0.2273,
+      "step": 58800
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.738651202089254e-05,
+      "loss": 0.2267,
+      "step": 58900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7348106613411167e-05,
+      "loss": 0.2271,
+      "step": 59000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7309701205929793e-05,
+      "loss": 0.2259,
+      "step": 59100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7271295798448422e-05,
+      "loss": 0.2247,
+      "step": 59200
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7232890390967048e-05,
+      "loss": 0.2269,
+      "step": 59300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7194484983485674e-05,
+      "loss": 0.2232,
+      "step": 59400
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.71560795760043e-05,
+      "loss": 0.228,
+      "step": 59500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7117674168522926e-05,
+      "loss": 0.2253,
+      "step": 59600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7079268761041555e-05,
+      "loss": 0.2293,
+      "step": 59700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.704086335356018e-05,
+      "loss": 0.2262,
+      "step": 59800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.700284200015362e-05,
+      "loss": 0.224,
+      "step": 59900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6964436592672247e-05,
+      "loss": 0.2246,
+      "step": 60000
+    },
+    {
+      "epoch": 4.61,
+      "eval_cer": 1.8035,
+      "eval_gen_len": 12.70475,
+      "eval_loss": 0.23027491569519043,
+      "eval_runtime": 1391.0574,
+      "eval_samples_per_second": 71.888,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 9.4897,
+      "step": 60000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.692641523926569e-05,
+      "loss": 0.2247,
+      "step": 60100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6888009831784317e-05,
+      "loss": 0.2253,
+      "step": 60200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6849604424302943e-05,
+      "loss": 0.2275,
+      "step": 60300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.681119901682157e-05,
+      "loss": 0.2265,
+      "step": 60400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6772793609340198e-05,
+      "loss": 0.224,
+      "step": 60500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6734388201858824e-05,
+      "loss": 0.226,
+      "step": 60600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.669598279437745e-05,
+      "loss": 0.2246,
+      "step": 60700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6657577386896076e-05,
+      "loss": 0.2271,
+      "step": 60800
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.66191719794147e-05,
+      "loss": 0.2232,
+      "step": 60900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.658076657193333e-05,
+      "loss": 0.224,
+      "step": 61000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6542361164451957e-05,
+      "loss": 0.226,
+      "step": 61100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6503955756970583e-05,
+      "loss": 0.2225,
+      "step": 61200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6465934403564023e-05,
+      "loss": 0.2233,
+      "step": 61300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6427528996082652e-05,
+      "loss": 0.2249,
+      "step": 61400
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6389123588601278e-05,
+      "loss": 0.2255,
+      "step": 61500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6350718181119904e-05,
+      "loss": 0.2263,
+      "step": 61600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.631231277363853e-05,
+      "loss": 0.224,
+      "step": 61700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.627390736615716e-05,
+      "loss": 0.2232,
+      "step": 61800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6235501958675785e-05,
+      "loss": 0.2239,
+      "step": 61900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.619709655119441e-05,
+      "loss": 0.2232,
+      "step": 62000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6158691143713037e-05,
+      "loss": 0.2222,
+      "step": 62100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6120285736231663e-05,
+      "loss": 0.2209,
+      "step": 62200
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6081880328750292e-05,
+      "loss": 0.2243,
+      "step": 62300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6043474921268918e-05,
+      "loss": 0.2199,
+      "step": 62400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6005069513787544e-05,
+      "loss": 0.226,
+      "step": 62500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.596666410630617e-05,
+      "loss": 0.2248,
+      "step": 62600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5928258698824796e-05,
+      "loss": 0.2208,
+      "step": 62700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5889853291343425e-05,
+      "loss": 0.2229,
+      "step": 62800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585144788386205e-05,
+      "loss": 0.2208,
+      "step": 62900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5813042476380677e-05,
+      "loss": 0.2202,
+      "step": 63000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5774637068899303e-05,
+      "loss": 0.2201,
+      "step": 63100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5736231661417933e-05,
+      "loss": 0.2209,
+      "step": 63200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.569782625393656e-05,
+      "loss": 0.2209,
+      "step": 63300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5659420846455184e-05,
+      "loss": 0.2214,
+      "step": 63400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.562101543897381e-05,
+      "loss": 0.2204,
+      "step": 63500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5582994085567247e-05,
+      "loss": 0.2207,
+      "step": 63600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5544588678085873e-05,
+      "loss": 0.2222,
+      "step": 63700
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.55061832706045e-05,
+      "loss": 0.2209,
+      "step": 63800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5467777863123128e-05,
+      "loss": 0.2174,
+      "step": 63900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5429372455641754e-05,
+      "loss": 0.2219,
+      "step": 64000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.539096704816038e-05,
+      "loss": 0.2174,
+      "step": 64100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5352561640679006e-05,
+      "loss": 0.2208,
+      "step": 64200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5314156233197632e-05,
+      "loss": 0.2227,
+      "step": 64300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.527575082571626e-05,
+      "loss": 0.2222,
+      "step": 64400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5237345418234887e-05,
+      "loss": 0.2219,
+      "step": 64500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5198940010753513e-05,
+      "loss": 0.2192,
+      "step": 64600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.516053460327214e-05,
+      "loss": 0.2177,
+      "step": 64700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5122129195790765e-05,
+      "loss": 0.2215,
+      "step": 64800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5083723788309394e-05,
+      "loss": 0.2195,
+      "step": 64900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.504531838082802e-05,
+      "loss": 0.2187,
+      "step": 65000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5006912973346646e-05,
+      "loss": 0.2161,
+      "step": 65100
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4968507565865276e-05,
+      "loss": 0.2006,
+      "step": 65200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.49301021583839e-05,
+      "loss": 0.2002,
+      "step": 65300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.489169675090253e-05,
+      "loss": 0.1976,
+      "step": 65400
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4853291343421157e-05,
+      "loss": 0.2011,
+      "step": 65500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4814885935939783e-05,
+      "loss": 0.2009,
+      "step": 65600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.477648052845841e-05,
+      "loss": 0.2024,
+      "step": 65700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4738075120977035e-05,
+      "loss": 0.2006,
+      "step": 65800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.469966971349566e-05,
+      "loss": 0.2023,
+      "step": 65900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4661264306014286e-05,
+      "loss": 0.2023,
+      "step": 66000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4622858898532912e-05,
+      "loss": 0.2005,
+      "step": 66100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.458445349105154e-05,
+      "loss": 0.201,
+      "step": 66200
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4546048083570168e-05,
+      "loss": 0.1966,
+      "step": 66300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4507642676088794e-05,
+      "loss": 0.2007,
+      "step": 66400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.446923726860742e-05,
+      "loss": 0.1994,
+      "step": 66500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.443121591520086e-05,
+      "loss": 0.2045,
+      "step": 66600
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.439281050771949e-05,
+      "loss": 0.2017,
+      "step": 66700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4354405100238115e-05,
+      "loss": 0.2023,
+      "step": 66800
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.431599969275674e-05,
+      "loss": 0.2027,
+      "step": 66900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4277594285275367e-05,
+      "loss": 0.2029,
+      "step": 67000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4239188877793996e-05,
+      "loss": 0.2019,
+      "step": 67100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4200783470312622e-05,
+      "loss": 0.2011,
+      "step": 67200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4162378062831248e-05,
+      "loss": 0.202,
+      "step": 67300
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4123972655349874e-05,
+      "loss": 0.2015,
+      "step": 67400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.40855672478685e-05,
+      "loss": 0.2004,
+      "step": 67500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.404716184038713e-05,
+      "loss": 0.2039,
+      "step": 67600
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4008756432905755e-05,
+      "loss": 0.2022,
+      "step": 67700
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.397035102542438e-05,
+      "loss": 0.2045,
+      "step": 67800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3931945617943007e-05,
+      "loss": 0.2018,
+      "step": 67900
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3893540210461633e-05,
+      "loss": 0.2018,
+      "step": 68000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3855134802980262e-05,
+      "loss": 0.2033,
+      "step": 68100
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3816729395498888e-05,
+      "loss": 0.2004,
+      "step": 68200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3778323988017514e-05,
+      "loss": 0.2008,
+      "step": 68300
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.373991858053614e-05,
+      "loss": 0.2015,
+      "step": 68400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.370151317305477e-05,
+      "loss": 0.206,
+      "step": 68500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3663107765573395e-05,
+      "loss": 0.2026,
+      "step": 68600
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.362470235809202e-05,
+      "loss": 0.2033,
+      "step": 68700
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3586296950610647e-05,
+      "loss": 0.2034,
+      "step": 68800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3547891543129273e-05,
+      "loss": 0.2035,
+      "step": 68900
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3509486135647902e-05,
+      "loss": 0.2015,
+      "step": 69000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.347108072816653e-05,
+      "loss": 0.2015,
+      "step": 69100
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3432675320685154e-05,
+      "loss": 0.2022,
+      "step": 69200
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.339426991320378e-05,
+      "loss": 0.1988,
+      "step": 69300
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.335586450572241e-05,
+      "loss": 0.1994,
+      "step": 69400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3317459098241036e-05,
+      "loss": 0.2026,
+      "step": 69500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.327905369075966e-05,
+      "loss": 0.2019,
+      "step": 69600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3240648283278287e-05,
+      "loss": 0.2038,
+      "step": 69700
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3202242875796913e-05,
+      "loss": 0.2001,
+      "step": 69800
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3163837468315543e-05,
+      "loss": 0.2024,
+      "step": 69900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.312543206083417e-05,
+      "loss": 0.2004,
+      "step": 70000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.308702665335279e-05,
+      "loss": 0.2056,
+      "step": 70100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3048621245871417e-05,
+      "loss": 0.2029,
+      "step": 70200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3010215838390046e-05,
+      "loss": 0.2029,
+      "step": 70300
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2971810430908672e-05,
+      "loss": 0.2033,
+      "step": 70400
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2933405023427298e-05,
+      "loss": 0.2,
+      "step": 70500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2894999615945924e-05,
+      "loss": 0.1987,
+      "step": 70600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.285659420846455e-05,
+      "loss": 0.2012,
+      "step": 70700
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.281818880098318e-05,
+      "loss": 0.2039,
+      "step": 70800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2779783393501805e-05,
+      "loss": 0.2044,
+      "step": 70900
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.274137798602043e-05,
+      "loss": 0.1989,
+      "step": 71000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2702972578539057e-05,
+      "loss": 0.202,
+      "step": 71100
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2664567171057687e-05,
+      "loss": 0.2024,
+      "step": 71200
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2626161763576313e-05,
+      "loss": 0.2032,
+      "step": 71300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.258775635609494e-05,
+      "loss": 0.2007,
+      "step": 71400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2549350948613564e-05,
+      "loss": 0.2007,
+      "step": 71500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.251094554113219e-05,
+      "loss": 0.203,
+      "step": 71600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.247254013365082e-05,
+      "loss": 0.2011,
+      "step": 71700
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2434134726169446e-05,
+      "loss": 0.1968,
+      "step": 71800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.239572931868807e-05,
+      "loss": 0.2016,
+      "step": 71900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2357323911206697e-05,
+      "loss": 0.2017,
+      "step": 72000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2318918503725327e-05,
+      "loss": 0.2011,
+      "step": 72100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2280513096243953e-05,
+      "loss": 0.2001,
+      "step": 72200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.224210768876258e-05,
+      "loss": 0.1985,
+      "step": 72300
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2203702281281205e-05,
+      "loss": 0.2008,
+      "step": 72400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.216529687379983e-05,
+      "loss": 0.204,
+      "step": 72500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.212689146631846e-05,
+      "loss": 0.2013,
+      "step": 72600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.20888701129119e-05,
+      "loss": 0.2035,
+      "step": 72700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2050464705430526e-05,
+      "loss": 0.2042,
+      "step": 72800
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2012059297949152e-05,
+      "loss": 0.201,
+      "step": 72900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1974037944542592e-05,
+      "loss": 0.2012,
+      "step": 73000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1935632537061218e-05,
+      "loss": 0.198,
+      "step": 73100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1897227129579844e-05,
+      "loss": 0.2002,
+      "step": 73200
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.185882172209847e-05,
+      "loss": 0.2007,
+      "step": 73300
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.18204163146171e-05,
+      "loss": 0.1989,
+      "step": 73400
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1782010907135725e-05,
+      "loss": 0.2026,
+      "step": 73500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.174360549965435e-05,
+      "loss": 0.2014,
+      "step": 73600
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1705200092172977e-05,
+      "loss": 0.1989,
+      "step": 73700
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1666794684691606e-05,
+      "loss": 0.1989,
+      "step": 73800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1628389277210232e-05,
+      "loss": 0.2,
+      "step": 73900
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1589983869728858e-05,
+      "loss": 0.1985,
+      "step": 74000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1551578462247484e-05,
+      "loss": 0.2007,
+      "step": 74100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.151317305476611e-05,
+      "loss": 0.2006,
+      "step": 74200
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.147476764728474e-05,
+      "loss": 0.1995,
+      "step": 74300
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1436362239803365e-05,
+      "loss": 0.1969,
+      "step": 74400
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.139795683232199e-05,
+      "loss": 0.1982,
+      "step": 74500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1359551424840617e-05,
+      "loss": 0.2022,
+      "step": 74600
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1321146017359246e-05,
+      "loss": 0.2025,
+      "step": 74700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1282740609877872e-05,
+      "loss": 0.2023,
+      "step": 74800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1244335202396498e-05,
+      "loss": 0.2002,
+      "step": 74900
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1205929794915124e-05,
+      "loss": 0.1998,
+      "step": 75000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.116752438743375e-05,
+      "loss": 0.2026,
+      "step": 75100
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.112911897995238e-05,
+      "loss": 0.1993,
+      "step": 75200
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1090713572471005e-05,
+      "loss": 0.1985,
+      "step": 75300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.105230816498963e-05,
+      "loss": 0.1996,
+      "step": 75400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1013902757508257e-05,
+      "loss": 0.1986,
+      "step": 75500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0975497350026883e-05,
+      "loss": 0.2015,
+      "step": 75600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0937091942545513e-05,
+      "loss": 0.198,
+      "step": 75700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.089868653506414e-05,
+      "loss": 0.198,
+      "step": 75800
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0860281127582764e-05,
+      "loss": 0.1984,
+      "step": 75900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.082187572010139e-05,
+      "loss": 0.1994,
+      "step": 76000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.078347031262002e-05,
+      "loss": 0.2011,
+      "step": 76100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0745064905138646e-05,
+      "loss": 0.2004,
+      "step": 76200
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.070665949765727e-05,
+      "loss": 0.1987,
+      "step": 76300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0668254090175897e-05,
+      "loss": 0.1999,
+      "step": 76400
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0629848682694523e-05,
+      "loss": 0.1973,
+      "step": 76500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0591443275213153e-05,
+      "loss": 0.1992,
+      "step": 76600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0553421921806593e-05,
+      "loss": 0.1975,
+      "step": 76700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.051501651432522e-05,
+      "loss": 0.199,
+      "step": 76800
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0476611106843845e-05,
+      "loss": 0.2009,
+      "step": 76900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.043820569936247e-05,
+      "loss": 0.2022,
+      "step": 77000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0399800291881097e-05,
+      "loss": 0.1972,
+      "step": 77100
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0361394884399723e-05,
+      "loss": 0.2003,
+      "step": 77200
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.032298947691835e-05,
+      "loss": 0.1986,
+      "step": 77300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0284584069436978e-05,
+      "loss": 0.1991,
+      "step": 77400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0246178661955604e-05,
+      "loss": 0.1987,
+      "step": 77500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.020777325447423e-05,
+      "loss": 0.1974,
+      "step": 77600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0169367846992856e-05,
+      "loss": 0.1993,
+      "step": 77700
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.013096243951148e-05,
+      "loss": 0.2001,
+      "step": 77800
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.009255703203011e-05,
+      "loss": 0.198,
+      "step": 77900
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0054151624548737e-05,
+      "loss": 0.1993,
+      "step": 78000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0015746217067363e-05,
+      "loss": 0.1991,
+      "step": 78100
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.997734080958599e-05,
+      "loss": 0.1842,
+      "step": 78200
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9938935402104618e-05,
+      "loss": 0.1815,
+      "step": 78300
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9900529994623244e-05,
+      "loss": 0.1827,
+      "step": 78400
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.986212458714187e-05,
+      "loss": 0.1814,
+      "step": 78500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9823719179660496e-05,
+      "loss": 0.1825,
+      "step": 78600
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9785313772179122e-05,
+      "loss": 0.1792,
+      "step": 78700
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.974690836469775e-05,
+      "loss": 0.1805,
+      "step": 78800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9708502957216377e-05,
+      "loss": 0.1833,
+      "step": 78900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9670097549735003e-05,
+      "loss": 0.1821,
+      "step": 79000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.963169214225363e-05,
+      "loss": 0.183,
+      "step": 79100
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9593286734772258e-05,
+      "loss": 0.1808,
+      "step": 79200
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9554881327290884e-05,
+      "loss": 0.1819,
+      "step": 79300
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.951647591980951e-05,
+      "loss": 0.1828,
+      "step": 79400
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9478070512328136e-05,
+      "loss": 0.1847,
+      "step": 79500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9439665104846762e-05,
+      "loss": 0.1828,
+      "step": 79600
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.940125969736539e-05,
+      "loss": 0.1816,
+      "step": 79700
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9362854289884017e-05,
+      "loss": 0.1828,
+      "step": 79800
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9324448882402643e-05,
+      "loss": 0.1825,
+      "step": 79900
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.928604347492127e-05,
+      "loss": 0.1812,
+      "step": 80000
+    },
+    {
+      "epoch": 6.14,
+      "eval_cer": 1.6804,
+      "eval_gen_len": 12.69369,
+      "eval_loss": 0.2121695876121521,
+      "eval_runtime": 1392.7952,
+      "eval_samples_per_second": 71.798,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 8.9349,
+      "step": 80000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9247638067439895e-05,
+      "loss": 0.1823,
+      "step": 80100
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9209232659958524e-05,
+      "loss": 0.184,
+      "step": 80200
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.917082725247715e-05,
+      "loss": 0.1824,
+      "step": 80300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9132421844995776e-05,
+      "loss": 0.1819,
+      "step": 80400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9094016437514402e-05,
+      "loss": 0.185,
+      "step": 80500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.905561103003303e-05,
+      "loss": 0.1834,
+      "step": 80600
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9017205622551657e-05,
+      "loss": 0.1848,
+      "step": 80700
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8978800215070283e-05,
+      "loss": 0.183,
+      "step": 80800
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.894039480758891e-05,
+      "loss": 0.1824,
+      "step": 80900
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8901989400107535e-05,
+      "loss": 0.1833,
+      "step": 81000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.886396804670098e-05,
+      "loss": 0.181,
+      "step": 81100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8825562639219605e-05,
+      "loss": 0.1832,
+      "step": 81200
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8787157231738227e-05,
+      "loss": 0.183,
+      "step": 81300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8748751824256857e-05,
+      "loss": 0.1812,
+      "step": 81400
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8710346416775482e-05,
+      "loss": 0.1821,
+      "step": 81500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.867194100929411e-05,
+      "loss": 0.1791,
+      "step": 81600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8633535601812734e-05,
+      "loss": 0.1805,
+      "step": 81700
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.859513019433136e-05,
+      "loss": 0.1853,
+      "step": 81800
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.855672478684999e-05,
+      "loss": 0.1842,
+      "step": 81900
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8518319379368616e-05,
+      "loss": 0.1819,
+      "step": 82000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.847991397188724e-05,
+      "loss": 0.1844,
+      "step": 82100
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8441508564405867e-05,
+      "loss": 0.1819,
+      "step": 82200
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8403103156924493e-05,
+      "loss": 0.1823,
+      "step": 82300
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8364697749443123e-05,
+      "loss": 0.1827,
+      "step": 82400
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.832629234196175e-05,
+      "loss": 0.1822,
+      "step": 82500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8287886934480375e-05,
+      "loss": 0.1815,
+      "step": 82600
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8249481526999e-05,
+      "loss": 0.1836,
+      "step": 82700
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8211460173592444e-05,
+      "loss": 0.182,
+      "step": 82800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8173438820185884e-05,
+      "loss": 0.1839,
+      "step": 82900
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.813503341270451e-05,
+      "loss": 0.1855,
+      "step": 83000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8096628005223136e-05,
+      "loss": 0.1832,
+      "step": 83100
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8058222597741765e-05,
+      "loss": 0.1849,
+      "step": 83200
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.801981719026039e-05,
+      "loss": 0.1842,
+      "step": 83300
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7981411782779017e-05,
+      "loss": 0.1822,
+      "step": 83400
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7943006375297643e-05,
+      "loss": 0.1821,
+      "step": 83500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7904600967816272e-05,
+      "loss": 0.1865,
+      "step": 83600
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.78661955603349e-05,
+      "loss": 0.1828,
+      "step": 83700
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.782779015285352e-05,
+      "loss": 0.1829,
+      "step": 83800
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7789384745372147e-05,
+      "loss": 0.1837,
+      "step": 83900
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7750979337890776e-05,
+      "loss": 0.1824,
+      "step": 84000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7712573930409402e-05,
+      "loss": 0.183,
+      "step": 84100
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7674168522928028e-05,
+      "loss": 0.1822,
+      "step": 84200
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7635763115446654e-05,
+      "loss": 0.1826,
+      "step": 84300
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.759735770796528e-05,
+      "loss": 0.1822,
+      "step": 84400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.755895230048391e-05,
+      "loss": 0.1838,
+      "step": 84500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7520546893002535e-05,
+      "loss": 0.1822,
+      "step": 84600
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.748214148552116e-05,
+      "loss": 0.1835,
+      "step": 84700
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7443736078039787e-05,
+      "loss": 0.1829,
+      "step": 84800
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7405330670558413e-05,
+      "loss": 0.1829,
+      "step": 84900
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7366925263077042e-05,
+      "loss": 0.1809,
+      "step": 85000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7328519855595668e-05,
+      "loss": 0.187,
+      "step": 85100
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7290114448114294e-05,
+      "loss": 0.1818,
+      "step": 85200
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.725170904063292e-05,
+      "loss": 0.1801,
+      "step": 85300
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.721330363315155e-05,
+      "loss": 0.1851,
+      "step": 85400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7174898225670175e-05,
+      "loss": 0.1828,
+      "step": 85500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.71364928181888e-05,
+      "loss": 0.1816,
+      "step": 85600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7098087410707427e-05,
+      "loss": 0.185,
+      "step": 85700
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7059682003226053e-05,
+      "loss": 0.1828,
+      "step": 85800
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 0.183,
+      "step": 85900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.698287118826331e-05,
+      "loss": 0.1847,
+      "step": 86000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6944465780781934e-05,
+      "loss": 0.1832,
+      "step": 86100
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.690606037330056e-05,
+      "loss": 0.1808,
+      "step": 86200
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.686765496581919e-05,
+      "loss": 0.1834,
+      "step": 86300
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6829249558337816e-05,
+      "loss": 0.1817,
+      "step": 86400
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.679084415085644e-05,
+      "loss": 0.1818,
+      "step": 86500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6752438743375067e-05,
+      "loss": 0.184,
+      "step": 86600
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6714033335893693e-05,
+      "loss": 0.1811,
+      "step": 86700
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6675627928412323e-05,
+      "loss": 0.1805,
+      "step": 86800
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.663722252093095e-05,
+      "loss": 0.1834,
+      "step": 86900
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6598817113449575e-05,
+      "loss": 0.1824,
+      "step": 87000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.65604117059682e-05,
+      "loss": 0.1797,
+      "step": 87100
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6522006298486826e-05,
+      "loss": 0.1807,
+      "step": 87200
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6483600891005456e-05,
+      "loss": 0.1823,
+      "step": 87300
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.644519548352408e-05,
+      "loss": 0.1834,
+      "step": 87400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6406790076042708e-05,
+      "loss": 0.1827,
+      "step": 87500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6368384668561334e-05,
+      "loss": 0.1803,
+      "step": 87600
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6329979261079963e-05,
+      "loss": 0.1823,
+      "step": 87700
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.629157385359859e-05,
+      "loss": 0.1816,
+      "step": 87800
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6253168446117215e-05,
+      "loss": 0.1799,
+      "step": 87900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.621476303863584e-05,
+      "loss": 0.1812,
+      "step": 88000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6176357631154467e-05,
+      "loss": 0.1789,
+      "step": 88100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6137952223673096e-05,
+      "loss": 0.1825,
+      "step": 88200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6099546816191722e-05,
+      "loss": 0.1842,
+      "step": 88300
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6061141408710348e-05,
+      "loss": 0.1801,
+      "step": 88400
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6022736001228974e-05,
+      "loss": 0.1795,
+      "step": 88500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.59843305937476e-05,
+      "loss": 0.1828,
+      "step": 88600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.594592518626623e-05,
+      "loss": 0.1817,
+      "step": 88700
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5907519778784855e-05,
+      "loss": 0.1811,
+      "step": 88800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.586949842537829e-05,
+      "loss": 0.1841,
+      "step": 88900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.583109301789692e-05,
+      "loss": 0.1788,
+      "step": 89000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5792687610415547e-05,
+      "loss": 0.1802,
+      "step": 89100
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5754282202934173e-05,
+      "loss": 0.1816,
+      "step": 89200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.57158767954528e-05,
+      "loss": 0.1818,
+      "step": 89300
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5677471387971425e-05,
+      "loss": 0.1804,
+      "step": 89400
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5639065980490054e-05,
+      "loss": 0.1791,
+      "step": 89500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.560066057300868e-05,
+      "loss": 0.1808,
+      "step": 89600
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5562255165527306e-05,
+      "loss": 0.1811,
+      "step": 89700
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5523849758045932e-05,
+      "loss": 0.1827,
+      "step": 89800
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.548544435056456e-05,
+      "loss": 0.1831,
+      "step": 89900
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5447038943083187e-05,
+      "loss": 0.1808,
+      "step": 90000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5408633535601813e-05,
+      "loss": 0.1817,
+      "step": 90100
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.537022812812044e-05,
+      "loss": 0.1822,
+      "step": 90200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5331822720639065e-05,
+      "loss": 0.1803,
+      "step": 90300
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5293417313157694e-05,
+      "loss": 0.181,
+      "step": 90400
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.525501190567632e-05,
+      "loss": 0.1786,
+      "step": 90500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5216606498194946e-05,
+      "loss": 0.1822,
+      "step": 90600
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5178201090713574e-05,
+      "loss": 0.1793,
+      "step": 90700
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.51397956832322e-05,
+      "loss": 0.181,
+      "step": 90800
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5101390275750826e-05,
+      "loss": 0.1825,
+      "step": 90900
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5062984868269453e-05,
+      "loss": 0.1822,
+      "step": 91000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5024963514862895e-05,
+      "loss": 0.1809,
+      "step": 91100
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4986558107381521e-05,
+      "loss": 0.174,
+      "step": 91200
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4948152699900147e-05,
+      "loss": 0.1668,
+      "step": 91300
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4909747292418775e-05,
+      "loss": 0.1688,
+      "step": 91400
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.48713418849374e-05,
+      "loss": 0.1682,
+      "step": 91500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4832936477456028e-05,
+      "loss": 0.1652,
+      "step": 91600
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4794531069974654e-05,
+      "loss": 0.1633,
+      "step": 91700
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4756125662493282e-05,
+      "loss": 0.1675,
+      "step": 91800
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4717720255011908e-05,
+      "loss": 0.1671,
+      "step": 91900
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4679314847530534e-05,
+      "loss": 0.1659,
+      "step": 92000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4640909440049161e-05,
+      "loss": 0.1676,
+      "step": 92100
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4602504032567787e-05,
+      "loss": 0.1668,
+      "step": 92200
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4564098625086415e-05,
+      "loss": 0.1654,
+      "step": 92300
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.452569321760504e-05,
+      "loss": 0.1675,
+      "step": 92400
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4487287810123665e-05,
+      "loss": 0.1697,
+      "step": 92500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4448882402642291e-05,
+      "loss": 0.1663,
+      "step": 92600
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4410476995160918e-05,
+      "loss": 0.1655,
+      "step": 92700
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4372071587679544e-05,
+      "loss": 0.1705,
+      "step": 92800
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4333666180198172e-05,
+      "loss": 0.1672,
+      "step": 92900
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4295260772716798e-05,
+      "loss": 0.1672,
+      "step": 93000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.425723941931024e-05,
+      "loss": 0.1682,
+      "step": 93100
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4218834011828866e-05,
+      "loss": 0.1668,
+      "step": 93200
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4180428604347493e-05,
+      "loss": 0.1669,
+      "step": 93300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.414202319686612e-05,
+      "loss": 0.1658,
+      "step": 93400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4103617789384745e-05,
+      "loss": 0.1682,
+      "step": 93500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4065212381903373e-05,
+      "loss": 0.1688,
+      "step": 93600
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4026806974421999e-05,
+      "loss": 0.1685,
+      "step": 93700
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.3988401566940626e-05,
+      "loss": 0.1673,
+      "step": 93800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3949996159459252e-05,
+      "loss": 0.1675,
+      "step": 93900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.391159075197788e-05,
+      "loss": 0.169,
+      "step": 94000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3873185344496506e-05,
+      "loss": 0.1682,
+      "step": 94100
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3834779937015132e-05,
+      "loss": 0.169,
+      "step": 94200
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.379637452953376e-05,
+      "loss": 0.1699,
+      "step": 94300
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3757969122052385e-05,
+      "loss": 0.1702,
+      "step": 94400
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3719563714571013e-05,
+      "loss": 0.1678,
+      "step": 94500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3681158307089639e-05,
+      "loss": 0.1691,
+      "step": 94600
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3642752899608267e-05,
+      "loss": 0.1669,
+      "step": 94700
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3604347492126893e-05,
+      "loss": 0.167,
+      "step": 94800
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3565942084645518e-05,
+      "loss": 0.1702,
+      "step": 94900
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3527536677164146e-05,
+      "loss": 0.1672,
+      "step": 95000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3489131269682772e-05,
+      "loss": 0.167,
+      "step": 95100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.34507258622014e-05,
+      "loss": 0.1694,
+      "step": 95200
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3412320454720026e-05,
+      "loss": 0.1679,
+      "step": 95300
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.3374299101313464e-05,
+      "loss": 0.1698,
+      "step": 95400
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3335893693832092e-05,
+      "loss": 0.1694,
+      "step": 95500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3297488286350718e-05,
+      "loss": 0.1669,
+      "step": 95600
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3259082878869345e-05,
+      "loss": 0.168,
+      "step": 95700
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3220677471387971e-05,
+      "loss": 0.1682,
+      "step": 95800
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3182272063906597e-05,
+      "loss": 0.1692,
+      "step": 95900
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3143866656425225e-05,
+      "loss": 0.1695,
+      "step": 96000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.310546124894385e-05,
+      "loss": 0.1673,
+      "step": 96100
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3067055841462478e-05,
+      "loss": 0.1714,
+      "step": 96200
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3028650433981104e-05,
+      "loss": 0.1668,
+      "step": 96300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.2990245026499732e-05,
+      "loss": 0.1684,
+      "step": 96400
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2951839619018358e-05,
+      "loss": 0.1695,
+      "step": 96500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2913434211536984e-05,
+      "loss": 0.167,
+      "step": 96600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2875028804055611e-05,
+      "loss": 0.1668,
+      "step": 96700
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2836623396574237e-05,
+      "loss": 0.1694,
+      "step": 96800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2798217989092865e-05,
+      "loss": 0.1716,
+      "step": 96900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2759812581611491e-05,
+      "loss": 0.1677,
+      "step": 97000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2721407174130119e-05,
+      "loss": 0.1712,
+      "step": 97100
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2683001766648744e-05,
+      "loss": 0.1697,
+      "step": 97200
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.264459635916737e-05,
+      "loss": 0.1684,
+      "step": 97300
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2606190951685998e-05,
+      "loss": 0.168,
+      "step": 97400
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2568169598279438e-05,
+      "loss": 0.1687,
+      "step": 97500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2529764190798066e-05,
+      "loss": 0.1675,
+      "step": 97600
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2491358783316692e-05,
+      "loss": 0.1687,
+      "step": 97700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2452953375835318e-05,
+      "loss": 0.1672,
+      "step": 97800
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2414547968353944e-05,
+      "loss": 0.1714,
+      "step": 97900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2376142560872571e-05,
+      "loss": 0.1679,
+      "step": 98000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2337737153391197e-05,
+      "loss": 0.1696,
+      "step": 98100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2299715799984639e-05,
+      "loss": 0.171,
+      "step": 98200
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2261310392503265e-05,
+      "loss": 0.169,
+      "step": 98300
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2222904985021893e-05,
+      "loss": 0.1682,
+      "step": 98400
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2184499577540518e-05,
+      "loss": 0.1684,
+      "step": 98500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2146094170059146e-05,
+      "loss": 0.17,
+      "step": 98600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2107688762577772e-05,
+      "loss": 0.1667,
+      "step": 98700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.20692833550964e-05,
+      "loss": 0.1689,
+      "step": 98800
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2030877947615026e-05,
+      "loss": 0.1645,
+      "step": 98900
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.1992472540133652e-05,
+      "loss": 0.1678,
+      "step": 99000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1954067132652277e-05,
+      "loss": 0.1692,
+      "step": 99100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1915661725170903e-05,
+      "loss": 0.1679,
+      "step": 99200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1877256317689531e-05,
+      "loss": 0.1664,
+      "step": 99300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1838850910208157e-05,
+      "loss": 0.1692,
+      "step": 99400
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1800445502726785e-05,
+      "loss": 0.1657,
+      "step": 99500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.176204009524541e-05,
+      "loss": 0.1666,
+      "step": 99600
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1723634687764038e-05,
+      "loss": 0.1685,
+      "step": 99700
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1685229280282664e-05,
+      "loss": 0.1674,
+      "step": 99800
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.164682387280129e-05,
+      "loss": 0.1687,
+      "step": 99900
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.1608418465319918e-05,
+      "loss": 0.1693,
+      "step": 100000
+    },
+    {
+      "epoch": 7.68,
+      "eval_cer": 1.589,
+      "eval_gen_len": 12.70446,
+      "eval_loss": 0.2001003473997116,
+      "eval_runtime": 1390.8956,
+      "eval_samples_per_second": 71.896,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 8.5464,
+      "step": 100000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1570013057838544e-05,
+      "loss": 0.1666,
+      "step": 100100
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1531607650357171e-05,
+      "loss": 0.1699,
+      "step": 100200
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1493202242875797e-05,
+      "loss": 0.168,
+      "step": 100300
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1454796835394425e-05,
+      "loss": 0.167,
+      "step": 100400
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.141639142791305e-05,
+      "loss": 0.1675,
+      "step": 100500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1377986020431677e-05,
+      "loss": 0.1685,
+      "step": 100600
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1339580612950304e-05,
+      "loss": 0.1671,
+      "step": 100700
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.130117520546893e-05,
+      "loss": 0.1696,
+      "step": 100800
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1262769797987558e-05,
+      "loss": 0.165,
+      "step": 100900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1224364390506184e-05,
+      "loss": 0.1691,
+      "step": 101000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1185958983024811e-05,
+      "loss": 0.1696,
+      "step": 101100
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1147553575543437e-05,
+      "loss": 0.1683,
+      "step": 101200
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1109148168062063e-05,
+      "loss": 0.1686,
+      "step": 101300
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1070742760580691e-05,
+      "loss": 0.1718,
+      "step": 101400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.1032337353099317e-05,
+      "loss": 0.1701,
+      "step": 101500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.0993931945617944e-05,
+      "loss": 0.1701,
+      "step": 101600
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0955910592211385e-05,
+      "loss": 0.1686,
+      "step": 101700
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.091750518473001e-05,
+      "loss": 0.1697,
+      "step": 101800
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0879099777248636e-05,
+      "loss": 0.1657,
+      "step": 101900
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0840694369767264e-05,
+      "loss": 0.1676,
+      "step": 102000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.080228896228589e-05,
+      "loss": 0.1673,
+      "step": 102100
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0763883554804518e-05,
+      "loss": 0.1678,
+      "step": 102200
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0725478147323144e-05,
+      "loss": 0.1688,
+      "step": 102300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0687072739841771e-05,
+      "loss": 0.1692,
+      "step": 102400
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0648667332360397e-05,
+      "loss": 0.1683,
+      "step": 102500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0610261924879023e-05,
+      "loss": 0.169,
+      "step": 102600
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.057185651739765e-05,
+      "loss": 0.1672,
+      "step": 102700
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0533451109916277e-05,
+      "loss": 0.1691,
+      "step": 102800
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0495045702434904e-05,
+      "loss": 0.1684,
+      "step": 102900
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.045664029495353e-05,
+      "loss": 0.1661,
+      "step": 103000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0418234887472158e-05,
+      "loss": 0.1651,
+      "step": 103100
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0379829479990784e-05,
+      "loss": 0.1683,
+      "step": 103200
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.034142407250941e-05,
+      "loss": 0.1644,
+      "step": 103300
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0303018665028036e-05,
+      "loss": 0.1668,
+      "step": 103400
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0264613257546663e-05,
+      "loss": 0.1634,
+      "step": 103500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.022620785006529e-05,
+      "loss": 0.1659,
+      "step": 103600
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0187802442583915e-05,
+      "loss": 0.167,
+      "step": 103700
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0149397035102543e-05,
+      "loss": 0.1656,
+      "step": 103800
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0110991627621169e-05,
+      "loss": 0.1679,
+      "step": 103900
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.007297027421461e-05,
+      "loss": 0.1687,
+      "step": 104000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0034564866733236e-05,
+      "loss": 0.1698,
+      "step": 104100
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.996159459251864e-06,
+      "loss": 0.1562,
+      "step": 104200
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.95775405177049e-06,
+      "loss": 0.1566,
+      "step": 104300
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.919348644289118e-06,
+      "loss": 0.1564,
+      "step": 104400
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.880943236807742e-06,
+      "loss": 0.1569,
+      "step": 104500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.84253782932637e-06,
+      "loss": 0.157,
+      "step": 104600
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.804132421844995e-06,
+      "loss": 0.1543,
+      "step": 104700
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.765727014363623e-06,
+      "loss": 0.1556,
+      "step": 104800
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.727321606882249e-06,
+      "loss": 0.1563,
+      "step": 104900
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.688916199400875e-06,
+      "loss": 0.1571,
+      "step": 105000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.650510791919503e-06,
+      "loss": 0.1553,
+      "step": 105100
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.612105384438129e-06,
+      "loss": 0.1578,
+      "step": 105200
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.573699976956756e-06,
+      "loss": 0.1549,
+      "step": 105300
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.535294569475382e-06,
+      "loss": 0.155,
+      "step": 105400
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.49688916199401e-06,
+      "loss": 0.1561,
+      "step": 105500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.458483754512636e-06,
+      "loss": 0.1572,
+      "step": 105600
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.420078347031262e-06,
+      "loss": 0.1588,
+      "step": 105700
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.38167293954989e-06,
+      "loss": 0.157,
+      "step": 105800
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.343267532068515e-06,
+      "loss": 0.157,
+      "step": 105900
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.304862124587143e-06,
+      "loss": 0.1574,
+      "step": 106000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.266456717105769e-06,
+      "loss": 0.1561,
+      "step": 106100
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.228051309624396e-06,
+      "loss": 0.1562,
+      "step": 106200
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.189645902143022e-06,
+      "loss": 0.1582,
+      "step": 106300
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.151240494661648e-06,
+      "loss": 0.1565,
+      "step": 106400
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.112835087180276e-06,
+      "loss": 0.1576,
+      "step": 106500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.074429679698902e-06,
+      "loss": 0.1574,
+      "step": 106600
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.03602427221753e-06,
+      "loss": 0.1566,
+      "step": 106700
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.997618864736155e-06,
+      "loss": 0.1575,
+      "step": 106800
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.959213457254783e-06,
+      "loss": 0.1581,
+      "step": 106900
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.920808049773409e-06,
+      "loss": 0.1549,
+      "step": 107000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.882402642292035e-06,
+      "loss": 0.1593,
+      "step": 107100
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.843997234810662e-06,
+      "loss": 0.1572,
+      "step": 107200
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.805591827329288e-06,
+      "loss": 0.16,
+      "step": 107300
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.767186419847916e-06,
+      "loss": 0.1567,
+      "step": 107400
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.728781012366542e-06,
+      "loss": 0.1579,
+      "step": 107500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.690375604885168e-06,
+      "loss": 0.1591,
+      "step": 107600
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.651970197403794e-06,
+      "loss": 0.1556,
+      "step": 107700
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.613564789922421e-06,
+      "loss": 0.1606,
+      "step": 107800
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.575159382441047e-06,
+      "loss": 0.1582,
+      "step": 107900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.53713802903449e-06,
+      "loss": 0.1572,
+      "step": 108000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.498732621553115e-06,
+      "loss": 0.1581,
+      "step": 108100
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.460327214071743e-06,
+      "loss": 0.1559,
+      "step": 108200
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.421921806590369e-06,
+      "loss": 0.1587,
+      "step": 108300
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.383516399108995e-06,
+      "loss": 0.1592,
+      "step": 108400
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.345110991627622e-06,
+      "loss": 0.1553,
+      "step": 108500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.306705584146248e-06,
+      "loss": 0.1578,
+      "step": 108600
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.268300176664876e-06,
+      "loss": 0.1595,
+      "step": 108700
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.229894769183502e-06,
+      "loss": 0.157,
+      "step": 108800
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.191489361702128e-06,
+      "loss": 0.1578,
+      "step": 108900
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.153083954220754e-06,
+      "loss": 0.1557,
+      "step": 109000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.114678546739381e-06,
+      "loss": 0.1576,
+      "step": 109100
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.076273139258007e-06,
+      "loss": 0.1571,
+      "step": 109200
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.037867731776635e-06,
+      "loss": 0.1597,
+      "step": 109300
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 7.99946232429526e-06,
+      "loss": 0.1572,
+      "step": 109400
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.961056916813887e-06,
+      "loss": 0.1566,
+      "step": 109500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.922651509332514e-06,
+      "loss": 0.1567,
+      "step": 109600
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.88424610185114e-06,
+      "loss": 0.1578,
+      "step": 109700
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.845840694369768e-06,
+      "loss": 0.156,
+      "step": 109800
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.807435286888394e-06,
+      "loss": 0.1551,
+      "step": 109900
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.769029879407021e-06,
+      "loss": 0.1549,
+      "step": 110000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.730624471925647e-06,
+      "loss": 0.158,
+      "step": 110100
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.692219064444273e-06,
+      "loss": 0.1596,
+      "step": 110200
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.654197711037713e-06,
+      "loss": 0.1587,
+      "step": 110300
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.61579230355634e-06,
+      "loss": 0.1566,
+      "step": 110400
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.577386896074967e-06,
+      "loss": 0.1574,
+      "step": 110500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.538981488593594e-06,
+      "loss": 0.1577,
+      "step": 110600
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.500576081112221e-06,
+      "loss": 0.1562,
+      "step": 110700
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.462170673630847e-06,
+      "loss": 0.1564,
+      "step": 110800
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.423765266149474e-06,
+      "loss": 0.1567,
+      "step": 110900
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.385359858668101e-06,
+      "loss": 0.1591,
+      "step": 111000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.346954451186727e-06,
+      "loss": 0.1591,
+      "step": 111100
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.308549043705354e-06,
+      "loss": 0.1586,
+      "step": 111200
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.2701436362239805e-06,
+      "loss": 0.1584,
+      "step": 111300
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.231738228742607e-06,
+      "loss": 0.1556,
+      "step": 111400
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.193332821261234e-06,
+      "loss": 0.1587,
+      "step": 111500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.154927413779861e-06,
+      "loss": 0.1592,
+      "step": 111600
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.116522006298488e-06,
+      "loss": 0.1557,
+      "step": 111700
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.0781165988171135e-06,
+      "loss": 0.1571,
+      "step": 111800
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.03971119133574e-06,
+      "loss": 0.1567,
+      "step": 111900
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.001305783854367e-06,
+      "loss": 0.1578,
+      "step": 112000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.962900376372994e-06,
+      "loss": 0.1543,
+      "step": 112100
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.924494968891621e-06,
+      "loss": 0.1585,
+      "step": 112200
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.8860895614102474e-06,
+      "loss": 0.1585,
+      "step": 112300
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.847684153928874e-06,
+      "loss": 0.1564,
+      "step": 112400
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.8096628005223135e-06,
+      "loss": 0.1573,
+      "step": 112500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.77125739304094e-06,
+      "loss": 0.158,
+      "step": 112600
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.732851985559567e-06,
+      "loss": 0.1554,
+      "step": 112700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.694446578078194e-06,
+      "loss": 0.1576,
+      "step": 112800
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.656041170596821e-06,
+      "loss": 0.1564,
+      "step": 112900
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.617635763115447e-06,
+      "loss": 0.1564,
+      "step": 113000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.579230355634073e-06,
+      "loss": 0.1592,
+      "step": 113100
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.5408249481527e-06,
+      "loss": 0.158,
+      "step": 113200
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.502419540671327e-06,
+      "loss": 0.1574,
+      "step": 113300
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.464014133189954e-06,
+      "loss": 0.1553,
+      "step": 113400
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.4256087257085805e-06,
+      "loss": 0.157,
+      "step": 113500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.387587372302021e-06,
+      "loss": 0.1568,
+      "step": 113600
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.3491819648206465e-06,
+      "loss": 0.157,
+      "step": 113700
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.310776557339273e-06,
+      "loss": 0.1558,
+      "step": 113800
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.2723711498579e-06,
+      "loss": 0.1583,
+      "step": 113900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.233965742376527e-06,
+      "loss": 0.1586,
+      "step": 114000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.195560334895154e-06,
+      "loss": 0.158,
+      "step": 114100
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.1571549274137804e-06,
+      "loss": 0.158,
+      "step": 114200
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.118749519932407e-06,
+      "loss": 0.1594,
+      "step": 114300
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.080344112451033e-06,
+      "loss": 0.1563,
+      "step": 114400
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.04193870496966e-06,
+      "loss": 0.1597,
+      "step": 114500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.003533297488287e-06,
+      "loss": 0.1584,
+      "step": 114600
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.9651278900069135e-06,
+      "loss": 0.1569,
+      "step": 114700
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.92672248252554e-06,
+      "loss": 0.1578,
+      "step": 114800
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.888317075044166e-06,
+      "loss": 0.158,
+      "step": 114900
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.849911667562793e-06,
+      "loss": 0.159,
+      "step": 115000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.81150626008142e-06,
+      "loss": 0.1556,
+      "step": 115100
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.7731008526000466e-06,
+      "loss": 0.1561,
+      "step": 115200
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.7346954451186725e-06,
+      "loss": 0.1579,
+      "step": 115300
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.696290037637299e-06,
+      "loss": 0.1574,
+      "step": 115400
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.657884630155926e-06,
+      "loss": 0.1567,
+      "step": 115500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.619479222674553e-06,
+      "loss": 0.1603,
+      "step": 115600
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.58107381519318e-06,
+      "loss": 0.1552,
+      "step": 115700
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.542668407711806e-06,
+      "loss": 0.1578,
+      "step": 115800
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.504263000230433e-06,
+      "loss": 0.1584,
+      "step": 115900
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.465857592749059e-06,
+      "loss": 0.16,
+      "step": 116000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.427452185267686e-06,
+      "loss": 0.1588,
+      "step": 116100
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.389046777786313e-06,
+      "loss": 0.1569,
+      "step": 116200
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.3506413703049395e-06,
+      "loss": 0.1584,
+      "step": 116300
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.312235962823566e-06,
+      "loss": 0.1563,
+      "step": 116400
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.273830555342192e-06,
+      "loss": 0.1566,
+      "step": 116500
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.235425147860819e-06,
+      "loss": 0.1563,
+      "step": 116600
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.197019740379446e-06,
+      "loss": 0.1583,
+      "step": 116700
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.158614332898072e-06,
+      "loss": 0.1572,
+      "step": 116800
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.1202089254166985e-06,
+      "loss": 0.1547,
+      "step": 116900
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.081803517935325e-06,
+      "loss": 0.1567,
+      "step": 117000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.043398110453952e-06,
+      "loss": 0.1557,
+      "step": 117100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.004992702972579e-06,
+      "loss": 0.1582,
+      "step": 117200
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.966587295491206e-06,
+      "loss": 0.1509,
+      "step": 117300
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.928181888009832e-06,
+      "loss": 0.1523,
+      "step": 117400
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.889776480528459e-06,
+      "loss": 0.1491,
+      "step": 117500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.851371073047085e-06,
+      "loss": 0.1466,
+      "step": 117600
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.812965665565712e-06,
+      "loss": 0.1494,
+      "step": 117700
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.774560258084339e-06,
+      "loss": 0.147,
+      "step": 117800
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.736538904677779e-06,
+      "loss": 0.1474,
+      "step": 117900
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.698133497196406e-06,
+      "loss": 0.1496,
+      "step": 118000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.659728089715032e-06,
+      "loss": 0.147,
+      "step": 118100
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.621322682233658e-06,
+      "loss": 0.1474,
+      "step": 118200
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.582917274752285e-06,
+      "loss": 0.1488,
+      "step": 118300
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.544511867270912e-06,
+      "loss": 0.1501,
+      "step": 118400
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.506106459789539e-06,
+      "loss": 0.1507,
+      "step": 118500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.4677010523081654e-06,
+      "loss": 0.1509,
+      "step": 118600
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.429295644826792e-06,
+      "loss": 0.1508,
+      "step": 118700
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.390890237345419e-06,
+      "loss": 0.1488,
+      "step": 118800
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.352484829864045e-06,
+      "loss": 0.151,
+      "step": 118900
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.314079422382672e-06,
+      "loss": 0.1504,
+      "step": 119000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.2756740149012985e-06,
+      "loss": 0.1479,
+      "step": 119100
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.237268607419925e-06,
+      "loss": 0.152,
+      "step": 119200
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.198863199938551e-06,
+      "loss": 0.1499,
+      "step": 119300
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.160841846531992e-06,
+      "loss": 0.1506,
+      "step": 119400
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.122436439050618e-06,
+      "loss": 0.1504,
+      "step": 119500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.084031031569245e-06,
+      "loss": 0.1532,
+      "step": 119600
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.045625624087872e-06,
+      "loss": 0.1504,
+      "step": 119700
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.0072202166064985e-06,
+      "loss": 0.1518,
+      "step": 119800
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.968814809125125e-06,
+      "loss": 0.1497,
+      "step": 119900
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.930409401643752e-06,
+      "loss": 0.1498,
+      "step": 120000
+    },
+    {
+      "epoch": 9.22,
+      "eval_cer": 1.5546,
+      "eval_gen_len": 12.69347,
+      "eval_loss": 0.19419625401496887,
+      "eval_runtime": 1387.0911,
+      "eval_samples_per_second": 72.093,
+      "eval_steps_per_second": 0.282,
+      "eval_wer": 8.3598,
+      "step": 120000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.892003994162379e-06,
+      "loss": 0.1503,
+      "step": 120100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.853598586681005e-06,
+      "loss": 0.1489,
+      "step": 120200
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.8151931791996315e-06,
+      "loss": 0.1484,
+      "step": 120300
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.7767877717182583e-06,
+      "loss": 0.1502,
+      "step": 120400
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.7383823642368842e-06,
+      "loss": 0.1474,
+      "step": 120500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.699976956755511e-06,
+      "loss": 0.1495,
+      "step": 120600
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.661571549274138e-06,
+      "loss": 0.1507,
+      "step": 120700
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6231661417927646e-06,
+      "loss": 0.1468,
+      "step": 120800
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.584760734311391e-06,
+      "loss": 0.1503,
+      "step": 120900
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.5463553268300177e-06,
+      "loss": 0.15,
+      "step": 121000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.5079499193486445e-06,
+      "loss": 0.1523,
+      "step": 121100
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.469544511867271e-06,
+      "loss": 0.1502,
+      "step": 121200
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4311391043858977e-06,
+      "loss": 0.1481,
+      "step": 121300
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.3927336969045244e-06,
+      "loss": 0.15,
+      "step": 121400
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.3543282894231512e-06,
+      "loss": 0.1498,
+      "step": 121500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.3159228819417776e-06,
+      "loss": 0.1516,
+      "step": 121600
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2775174744604044e-06,
+      "loss": 0.1493,
+      "step": 121700
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.239112066979031e-06,
+      "loss": 0.1495,
+      "step": 121800
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2007066594976575e-06,
+      "loss": 0.1492,
+      "step": 121900
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1623012520162843e-06,
+      "loss": 0.1492,
+      "step": 122000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.1238958445349106e-06,
+      "loss": 0.1519,
+      "step": 122100
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0854904370535374e-06,
+      "loss": 0.1512,
+      "step": 122200
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0470850295721642e-06,
+      "loss": 0.1493,
+      "step": 122300
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.0086796220907906e-06,
+      "loss": 0.1511,
+      "step": 122400
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.970274214609417e-06,
+      "loss": 0.1519,
+      "step": 122500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.9318688071280437e-06,
+      "loss": 0.1482,
+      "step": 122600
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8934633996466705e-06,
+      "loss": 0.1503,
+      "step": 122700
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.855057992165297e-06,
+      "loss": 0.1507,
+      "step": 122800
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8166525846839236e-06,
+      "loss": 0.1504,
+      "step": 122900
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.7782471772025504e-06,
+      "loss": 0.1513,
+      "step": 123000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.739841769721177e-06,
+      "loss": 0.1497,
+      "step": 123100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.701436362239803e-06,
+      "loss": 0.1496,
+      "step": 123200
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.66303095475843e-06,
+      "loss": 0.1502,
+      "step": 123300
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6246255472770567e-06,
+      "loss": 0.1471,
+      "step": 123400
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.5862201397956835e-06,
+      "loss": 0.1504,
+      "step": 123500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.54781473231431e-06,
+      "loss": 0.1499,
+      "step": 123600
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.5097933789077504e-06,
+      "loss": 0.15,
+      "step": 123700
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.4713879714263767e-06,
+      "loss": 0.1504,
+      "step": 123800
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4333666180198173e-06,
+      "loss": 0.1478,
+      "step": 123900
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.394961210538444e-06,
+      "loss": 0.1529,
+      "step": 124000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.356555803057071e-06,
+      "loss": 0.1508,
+      "step": 124100
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.318150395575697e-06,
+      "loss": 0.1517,
+      "step": 124200
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.2797449880943236e-06,
+      "loss": 0.1499,
+      "step": 124300
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2413395806129504e-06,
+      "loss": 0.1508,
+      "step": 124400
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.202934173131577e-06,
+      "loss": 0.1504,
+      "step": 124500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1645287656502035e-06,
+      "loss": 0.1496,
+      "step": 124600
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1261233581688303e-06,
+      "loss": 0.1491,
+      "step": 124700
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.087717950687457e-06,
+      "loss": 0.1501,
+      "step": 124800
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0493125432060834e-06,
+      "loss": 0.1478,
+      "step": 124900
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.01090713572471e-06,
+      "loss": 0.1518,
+      "step": 125000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.9725017282433366e-06,
+      "loss": 0.1476,
+      "step": 125100
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9340963207619633e-06,
+      "loss": 0.1481,
+      "step": 125200
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.89569091328059e-06,
+      "loss": 0.1486,
+      "step": 125300
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8572855057992167e-06,
+      "loss": 0.1503,
+      "step": 125400
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8188800983178433e-06,
+      "loss": 0.1512,
+      "step": 125500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.78047469083647e-06,
+      "loss": 0.1468,
+      "step": 125600
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7420692833550964e-06,
+      "loss": 0.1476,
+      "step": 125700
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.703663875873723e-06,
+      "loss": 0.1515,
+      "step": 125800
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.6652584683923496e-06,
+      "loss": 0.1491,
+      "step": 125900
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6268530609109763e-06,
+      "loss": 0.1487,
+      "step": 126000
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.588447653429603e-06,
+      "loss": 0.1497,
+      "step": 126100
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5500422459482297e-06,
+      "loss": 0.1513,
+      "step": 126200
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.511636838466856e-06,
+      "loss": 0.1488,
+      "step": 126300
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4732314309854828e-06,
+      "loss": 0.1475,
+      "step": 126400
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4348260235041094e-06,
+      "loss": 0.1506,
+      "step": 126500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3964206160227362e-06,
+      "loss": 0.1463,
+      "step": 126600
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3580152085413625e-06,
+      "loss": 0.1497,
+      "step": 126700
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3196098010599893e-06,
+      "loss": 0.1476,
+      "step": 126800
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.281204393578616e-06,
+      "loss": 0.1484,
+      "step": 126900
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2427989860972427e-06,
+      "loss": 0.1506,
+      "step": 127000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.204393578615869e-06,
+      "loss": 0.1491,
+      "step": 127100
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1659881711344958e-06,
+      "loss": 0.1489,
+      "step": 127200
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1275827636531224e-06,
+      "loss": 0.1497,
+      "step": 127300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0891773561717492e-06,
+      "loss": 0.1499,
+      "step": 127400
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0507719486903755e-06,
+      "loss": 0.1504,
+      "step": 127500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0123665412090023e-06,
+      "loss": 0.1489,
+      "step": 127600
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.739611337276289e-07,
+      "loss": 0.1486,
+      "step": 127700
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.355557262462555e-07,
+      "loss": 0.1482,
+      "step": 127800
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 8.971503187648821e-07,
+      "loss": 0.1499,
+      "step": 127900
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.587449112835088e-07,
+      "loss": 0.1507,
+      "step": 128000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.203395038021354e-07,
+      "loss": 0.1515,
+      "step": 128100
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.823181503955758e-07,
+      "loss": 0.1471,
+      "step": 128200
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.439127429142024e-07,
+      "loss": 0.1466,
+      "step": 128300
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.05507335432829e-07,
+      "loss": 0.149,
+      "step": 128400
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.671019279514556e-07,
+      "loss": 0.1479,
+      "step": 128500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.286965204700822e-07,
+      "loss": 0.1467,
+      "step": 128600
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.902911129887089e-07,
+      "loss": 0.1507,
+      "step": 128700
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.518857055073355e-07,
+      "loss": 0.1496,
+      "step": 128800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.134802980259621e-07,
+      "loss": 0.1472,
+      "step": 128900
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.750748905445887e-07,
+      "loss": 0.1505,
+      "step": 129000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.366694830632154e-07,
+      "loss": 0.1478,
+      "step": 129100
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.9826407558184195e-07,
+      "loss": 0.1482,
+      "step": 129200
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.5985866810046857e-07,
+      "loss": 0.1508,
+      "step": 129300
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.214532606190952e-07,
+      "loss": 0.1503,
+      "step": 129400
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.830478531377218e-07,
+      "loss": 0.1517,
+      "step": 129500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.4464244565634844e-07,
+      "loss": 0.1508,
+      "step": 129600
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.0623703817497504e-07,
+      "loss": 0.1505,
+      "step": 129700
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6783163069360166e-07,
+      "loss": 0.1506,
+      "step": 129800
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.2942622321222829e-07,
+      "loss": 0.1495,
+      "step": 129900
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 9.102081573085491e-08,
+      "loss": 0.1498,
+      "step": 130000
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.261540824948153e-08,
+      "loss": 0.1483,
+      "step": 130100
+    },
+    {
+      "epoch": 10.0,
+      "step": 130190,
+      "total_flos": 3.716902955630592e+17,
+      "train_loss": 0.3488236354274385,
+      "train_runtime": 35245.4345,
+      "train_samples_per_second": 945.549,
+      "train_steps_per_second": 3.694
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 130190,
+  "num_train_epochs": 10,
+  "save_steps": 10000,
+  "total_flos": 3.716902955630592e+17,
+  "trial_name": null,
+  "trial_params": null
+}