Training in progress, step 1000, checkpoint

last-checkpoint/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7866bebbaf9b8e517a84648e60803297129c3fd969715fbdeaae9b2db38d25b6
+oid sha256:c860a83fb5b9f69118a6d8d4b59a20a11f110515ffcdbc5cd3807e7b1b14b397
 size 223144592

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56b99d87026a8ade6094641933fad0d4a3c0915eefa6662c667db9b66f557d4e
+oid sha256:e99dd7e87e0550442caf560a5b4216f0dbb5499430735f63c030308d75e8148e
 size 440235130

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b83c8e2618548ad0a6cb55d5f9e419ee82c921eaf0cb178e6e4f4700b370bc6
+oid sha256:8c05de932ead88280d2522d175713d98e67fbc9cde4f0986f914945e9eeeae9b
 size 14244

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9b5c0812245fd57418b0ad79ee0dda344f24eb5c34a21966b50dcec7dccda08
+oid sha256:ac8b3aeaa223f49129197d9577ea3c7b67e3ed41a56eef76042b6d22ac2afeb7
 size 1064

last-checkpoint/trainer_state.json

@@ -1,2902 +1,302 @@
 {
-  "best_metric": Infinity,
-  "best_model_checkpoint": null,
-  "epoch": 16.50206440957886,
+  "best_metric": 99.67939651107967,
+  "best_model_checkpoint": "./iteboshi_temp/checkpoint-1000",
+  "epoch": 1.6507018992568125,
   "eval_steps": 1000,
-  "global_step": 10000,
+  "global_step": 1000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.04128819157720892,
-      "grad_norm": 48.47208786010742,
+      "grad_norm": 48.469512939453125,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 17.3809,
+      "loss": 17.381,
       "step": 25
     },
     {
       "epoch": 0.08257638315441784,
-      "grad_norm": 52.7760009765625,
+      "grad_norm": 52.77425003051758,
       "learning_rate": 1.9600000000000003e-06,
       "loss": 15.9672,
       "step": 50
     },
     {
       "epoch": 0.12386457473162675,
-      "grad_norm": 64.02676391601562,
+      "grad_norm": 63.97101974487305,
       "learning_rate": 2.96e-06,
-      "loss": 12.6576,
+      "loss": 12.6578,
       "step": 75
     },
     {
       "epoch": 0.16515276630883569,
-      "grad_norm": 26.37363052368164,
+      "grad_norm": 26.345178604125977,
       "learning_rate": 3.96e-06,
-      "loss": 8.293,
+      "loss": 8.2932,
       "step": 100
     },
     {
       "epoch": 0.20644095788604458,
-      "grad_norm": 12.368265151977539,
+      "grad_norm": 12.366632461547852,
       "learning_rate": 4.960000000000001e-06,
-      "loss": 6.2484,
+      "loss": 6.2485,
       "step": 125
     },
     {
       "epoch": 0.2477291494632535,
-      "grad_norm": 9.140127182006836,
+      "grad_norm": 9.131484985351562,
       "learning_rate": 5.9600000000000005e-06,
       "loss": 5.0845,
       "step": 150
     },
     {
       "epoch": 0.28901734104046245,
-      "grad_norm": 8.285837173461914,
+      "grad_norm": 8.288640975952148,
       "learning_rate": 6.96e-06,
-      "loss": 4.5217,
+      "loss": 4.5216,
       "step": 175
     },
     {
       "epoch": 0.33030553261767137,
-      "grad_norm": 8.110857009887695,
+      "grad_norm": 8.107129096984863,
       "learning_rate": 7.960000000000002e-06,
       "loss": 4.2152,
       "step": 200
     },
     {
       "epoch": 0.37159372419488024,
-      "grad_norm": 8.587676048278809,
+      "grad_norm": 8.564615249633789,
       "learning_rate": 8.96e-06,
       "loss": 4.095,
       "step": 225
     },
     {
       "epoch": 0.41288191577208916,
-      "grad_norm": 8.696348190307617,
+      "grad_norm": 8.67183780670166,
       "learning_rate": 9.960000000000001e-06,
-      "loss": 3.9265,
+      "loss": 3.9264,
       "step": 250
     },
     {
       "epoch": 0.4541701073492981,
-      "grad_norm": 8.708649635314941,
+      "grad_norm": 8.779706954956055,
       "learning_rate": 1.0960000000000002e-05,
-      "loss": 3.7976,
+      "loss": 3.7974,
       "step": 275
     },
     {
       "epoch": 0.495458298926507,
-      "grad_norm": 8.679261207580566,
+      "grad_norm": 8.62621021270752,
       "learning_rate": 1.196e-05,
-      "loss": 3.7627,
+      "loss": 3.7625,
       "step": 300
     },
     {
       "epoch": 0.5367464905037159,
-      "grad_norm": 9.209630012512207,
+      "grad_norm": 9.157522201538086,
       "learning_rate": 1.2960000000000001e-05,
-      "loss": 3.6967,
+      "loss": 3.6965,
       "step": 325
     },
     {
       "epoch": 0.5780346820809249,
-      "grad_norm": 8.386543273925781,
+      "grad_norm": 8.446877479553223,
       "learning_rate": 1.396e-05,
       "loss": 3.6216,
       "step": 350
     },
     {
       "epoch": 0.6193228736581338,
-      "grad_norm": 8.898401260375977,
+      "grad_norm": 8.873468399047852,
       "learning_rate": 1.496e-05,
-      "loss": 3.5456,
+      "loss": 3.5455,
       "step": 375
     },
     {
       "epoch": 0.6606110652353427,
-      "grad_norm": 7.772883415222168,
+      "grad_norm": 7.768847465515137,
       "learning_rate": 1.5960000000000003e-05,
-      "loss": 3.4954,
+      "loss": 3.4952,
       "step": 400
     },
     {
       "epoch": 0.7018992568125516,
-      "grad_norm": 8.508569717407227,
+      "grad_norm": 8.365340232849121,
       "learning_rate": 1.696e-05,
-      "loss": 3.4144,
+      "loss": 3.414,
       "step": 425
     },
     {
       "epoch": 0.7431874483897605,
-      "grad_norm": 7.907943248748779,
+      "grad_norm": 7.865358829498291,
       "learning_rate": 1.796e-05,
-      "loss": 3.3481,
+      "loss": 3.3472,
       "step": 450
     },
     {
       "epoch": 0.7844756399669695,
-      "grad_norm": 8.553926467895508,
+      "grad_norm": 8.56877613067627,
       "learning_rate": 1.896e-05,
-      "loss": 3.3021,
+      "loss": 3.3018,
       "step": 475
     },
     {
       "epoch": 0.8257638315441783,
-      "grad_norm": 9.033528327941895,
+      "grad_norm": 9.091480255126953,
       "learning_rate": 1.9960000000000002e-05,
-      "loss": 3.2553,
+      "loss": 3.255,
       "step": 500
     },
     {
       "epoch": 0.8670520231213873,
-      "grad_norm": 8.92698860168457,
+      "grad_norm": 8.974713325500488,
       "learning_rate": 1.9949473684210527e-05,
-      "loss": 3.1524,
+      "loss": 3.1519,
       "step": 525
     },
     {
       "epoch": 0.9083402146985962,
-      "grad_norm": 9.119793891906738,
+      "grad_norm": 9.134045600891113,
       "learning_rate": 1.989684210526316e-05,
-      "loss": 3.034,
+      "loss": 3.0334,
       "step": 550
     },
     {
       "epoch": 0.9496284062758051,
-      "grad_norm": 10.161808967590332,
+      "grad_norm": 10.075201988220215,
       "learning_rate": 1.984421052631579e-05,
-      "loss": 2.9658,
+      "loss": 2.9656,
       "step": 575
     },
     {
       "epoch": 0.990916597853014,
-      "grad_norm": 9.213955879211426,
+      "grad_norm": 9.26815128326416,
       "learning_rate": 1.9791578947368423e-05,
-      "loss": 2.8887,
+      "loss": 2.8884,
       "step": 600
     },
     {
       "epoch": 1.0313790255986788,
-      "grad_norm": 8.941823959350586,
+      "grad_norm": 8.885565757751465,
       "learning_rate": 1.9738947368421057e-05,
-      "loss": 2.719,
+      "loss": 2.7179,
       "step": 625
     },
     {
       "epoch": 1.0726672171758878,
-      "grad_norm": 8.994600296020508,
+      "grad_norm": 8.9259672164917,
       "learning_rate": 1.9686315789473687e-05,
-      "loss": 2.6422,
+      "loss": 2.6417,
       "step": 650
     },
     {
       "epoch": 1.1139554087530965,
-      "grad_norm": 10.292838096618652,
+      "grad_norm": 10.3223295211792,
       "learning_rate": 1.9633684210526316e-05,
-      "loss": 2.5674,
+      "loss": 2.5669,
       "step": 675
     },
     {
       "epoch": 1.1552436003303055,
-      "grad_norm": 10.515816688537598,
+      "grad_norm": 10.508142471313477,
       "learning_rate": 1.958105263157895e-05,
-      "loss": 2.5031,
+      "loss": 2.5026,
       "step": 700
     },
     {
       "epoch": 1.1965317919075145,
-      "grad_norm": 9.79793643951416,
+      "grad_norm": 9.78288745880127,
       "learning_rate": 1.952842105263158e-05,
-      "loss": 2.3914,
+      "loss": 2.3907,
       "step": 725
     },
     {
       "epoch": 1.2378199834847234,
-      "grad_norm": 9.563103675842285,
+      "grad_norm": 9.629091262817383,
       "learning_rate": 1.9475789473684213e-05,
-      "loss": 2.3986,
+      "loss": 2.3983,
       "step": 750
     },
     {
       "epoch": 1.2791081750619324,
-      "grad_norm": 9.516156196594238,
+      "grad_norm": 9.543179512023926,
       "learning_rate": 1.9423157894736843e-05,
-      "loss": 2.2832,
+      "loss": 2.283,
       "step": 775
     },
     {
       "epoch": 1.3203963666391412,
-      "grad_norm": 9.650139808654785,
+      "grad_norm": 9.676267623901367,
       "learning_rate": 1.9370526315789476e-05,
-      "loss": 2.252,
+      "loss": 2.2519,
       "step": 800
     },
     {
       "epoch": 1.3616845582163501,
-      "grad_norm": 9.165299415588379,
+      "grad_norm": 9.192870140075684,
       "learning_rate": 1.9317894736842106e-05,
-      "loss": 2.1517,
+      "loss": 2.1516,
       "step": 825
     },
     {
       "epoch": 1.402972749793559,
-      "grad_norm": 9.827438354492188,
+      "grad_norm": 9.859146118164062,
       "learning_rate": 1.926526315789474e-05,
       "loss": 2.1643,
       "step": 850
     },
     {
       "epoch": 1.4442609413707679,
-      "grad_norm": 9.822501182556152,
+      "grad_norm": 9.815247535705566,
       "learning_rate": 1.921263157894737e-05,
-      "loss": 2.0519,
+      "loss": 2.0514,
       "step": 875
     },
     {
       "epoch": 1.4855491329479769,
-      "grad_norm": 10.055537223815918,
+      "grad_norm": 10.037995338439941,
       "learning_rate": 1.916e-05,
-      "loss": 2.0872,
+      "loss": 2.0867,
       "step": 900
     },
     {
       "epoch": 1.5268373245251858,
-      "grad_norm": 9.642505645751953,
+      "grad_norm": 9.680682182312012,
       "learning_rate": 1.9107368421052633e-05,
       "loss": 1.999,
       "step": 925
     },
     {
       "epoch": 1.5681255161023948,
-      "grad_norm": 9.428027153015137,
+      "grad_norm": 9.410001754760742,
       "learning_rate": 1.9054736842105263e-05,
       "loss": 1.954,
       "step": 950
     },
     {
       "epoch": 1.6094137076796038,
-      "grad_norm": 8.909976959228516,
+      "grad_norm": 8.89158821105957,
       "learning_rate": 1.9002105263157896e-05,
-      "loss": 1.9201,
+      "loss": 1.92,
       "step": 975
     },
     {
       "epoch": 1.6507018992568125,
-      "grad_norm": 9.313642501831055,
+      "grad_norm": 9.326930046081543,
       "learning_rate": 1.894947368421053e-05,
-      "loss": 1.834,
+      "loss": 1.8339,
       "step": 1000
     },
     {
       "epoch": 1.6507018992568125,
-      "eval_loss": 1.9117295742034912,
-      "eval_runtime": 784.2462,
-      "eval_samples_per_second": 13.492,
-      "eval_steps_per_second": 1.125,
-      "eval_wer": Infinity,
+      "eval_cer": 93.62053351369322,
+      "eval_loss": 1.911484956741333,
+      "eval_runtime": 801.8096,
+      "eval_samples_per_second": 13.196,
+      "eval_steps_per_second": 1.1,
+      "eval_wer": 99.67939651107967,
       "step": 1000
-    },
-    {
-      "epoch": 1.6919900908340215,
-      "grad_norm": 9.633246421813965,
-      "learning_rate": 1.889684210526316e-05,
-      "loss": 1.8354,
-      "step": 1025
-    },
-    {
-      "epoch": 1.7332782824112303,
-      "grad_norm": 9.896079063415527,
-      "learning_rate": 1.8844210526315793e-05,
-      "loss": 1.7921,
-      "step": 1050
-    },
-    {
-      "epoch": 1.7745664739884393,
-      "grad_norm": 9.53538703918457,
-      "learning_rate": 1.8791578947368423e-05,
-      "loss": 1.7655,
-      "step": 1075
-    },
-    {
-      "epoch": 1.8158546655656482,
-      "grad_norm": 9.284989356994629,
-      "learning_rate": 1.8738947368421056e-05,
-      "loss": 1.7743,
-      "step": 1100
-    },
-    {
-      "epoch": 1.8571428571428572,
-      "grad_norm": 10.101766586303711,
-      "learning_rate": 1.8686315789473686e-05,
-      "loss": 1.7346,
-      "step": 1125
-    },
-    {
-      "epoch": 1.8984310487200662,
-      "grad_norm": 8.92200756072998,
-      "learning_rate": 1.8633684210526316e-05,
-      "loss": 1.6885,
-      "step": 1150
-    },
-    {
-      "epoch": 1.939719240297275,
-      "grad_norm": 9.907315254211426,
-      "learning_rate": 1.858105263157895e-05,
-      "loss": 1.6912,
-      "step": 1175
-    },
-    {
-      "epoch": 1.981007431874484,
-      "grad_norm": 9.401750564575195,
-      "learning_rate": 1.852842105263158e-05,
-      "loss": 1.6279,
-      "step": 1200
-    },
-    {
-      "epoch": 2.0214698596201486,
-      "grad_norm": 10.461494445800781,
-      "learning_rate": 1.8475789473684212e-05,
-      "loss": 1.5268,
-      "step": 1225
-    },
-    {
-      "epoch": 2.0627580511973576,
-      "grad_norm": 8.997374534606934,
-      "learning_rate": 1.8423157894736842e-05,
-      "loss": 1.4142,
-      "step": 1250
-    },
-    {
-      "epoch": 2.1040462427745665,
-      "grad_norm": 9.4984712600708,
-      "learning_rate": 1.8370526315789476e-05,
-      "loss": 1.4769,
-      "step": 1275
-    },
-    {
-      "epoch": 2.1453344343517755,
-      "grad_norm": 9.436307907104492,
-      "learning_rate": 1.831789473684211e-05,
-      "loss": 1.4284,
-      "step": 1300
-    },
-    {
-      "epoch": 2.1866226259289845,
-      "grad_norm": 9.324549674987793,
-      "learning_rate": 1.826526315789474e-05,
-      "loss": 1.4117,
-      "step": 1325
-    },
-    {
-      "epoch": 2.227910817506193,
-      "grad_norm": 9.608540534973145,
-      "learning_rate": 1.821263157894737e-05,
-      "loss": 1.4093,
-      "step": 1350
-    },
-    {
-      "epoch": 2.269199009083402,
-      "grad_norm": 8.661190032958984,
-      "learning_rate": 1.8160000000000002e-05,
-      "loss": 1.3663,
-      "step": 1375
-    },
-    {
-      "epoch": 2.310487200660611,
-      "grad_norm": 9.19803524017334,
-      "learning_rate": 1.8107368421052632e-05,
-      "loss": 1.3647,
-      "step": 1400
-    },
-    {
-      "epoch": 2.35177539223782,
-      "grad_norm": 9.229018211364746,
-      "learning_rate": 1.8054736842105266e-05,
-      "loss": 1.3282,
-      "step": 1425
-    },
-    {
-      "epoch": 2.393063583815029,
-      "grad_norm": 9.667059898376465,
-      "learning_rate": 1.8002105263157896e-05,
-      "loss": 1.3098,
-      "step": 1450
-    },
-    {
-      "epoch": 2.434351775392238,
-      "grad_norm": 8.727728843688965,
-      "learning_rate": 1.794947368421053e-05,
-      "loss": 1.3246,
-      "step": 1475
-    },
-    {
-      "epoch": 2.475639966969447,
-      "grad_norm": 9.386205673217773,
-      "learning_rate": 1.789684210526316e-05,
-      "loss": 1.2703,
-      "step": 1500
-    },
-    {
-      "epoch": 2.516928158546656,
-      "grad_norm": 9.435502052307129,
-      "learning_rate": 1.7844210526315792e-05,
-      "loss": 1.3239,
-      "step": 1525
-    },
-    {
-      "epoch": 2.558216350123865,
-      "grad_norm": 8.652278900146484,
-      "learning_rate": 1.7791578947368422e-05,
-      "loss": 1.2833,
-      "step": 1550
-    },
-    {
-      "epoch": 2.5995045417010734,
-      "grad_norm": 8.806912422180176,
-      "learning_rate": 1.7738947368421052e-05,
-      "loss": 1.2917,
-      "step": 1575
-    },
-    {
-      "epoch": 2.6407927332782823,
-      "grad_norm": 9.386215209960938,
-      "learning_rate": 1.7686315789473685e-05,
-      "loss": 1.2775,
-      "step": 1600
-    },
-    {
-      "epoch": 2.6820809248554913,
-      "grad_norm": 8.931626319885254,
-      "learning_rate": 1.7633684210526315e-05,
-      "loss": 1.2501,
-      "step": 1625
-    },
-    {
-      "epoch": 2.7233691164327003,
-      "grad_norm": 8.861407279968262,
-      "learning_rate": 1.758105263157895e-05,
-      "loss": 1.2031,
-      "step": 1650
-    },
-    {
-      "epoch": 2.7646573080099093,
-      "grad_norm": 8.485048294067383,
-      "learning_rate": 1.7528421052631582e-05,
-      "loss": 1.2482,
-      "step": 1675
-    },
-    {
-      "epoch": 2.805945499587118,
-      "grad_norm": 9.203420639038086,
-      "learning_rate": 1.7475789473684212e-05,
-      "loss": 1.2105,
-      "step": 1700
-    },
-    {
-      "epoch": 2.847233691164327,
-      "grad_norm": 9.27082633972168,
-      "learning_rate": 1.7423157894736845e-05,
-      "loss": 1.2441,
-      "step": 1725
-    },
-    {
-      "epoch": 2.8885218827415358,
-      "grad_norm": 9.1829833984375,
-      "learning_rate": 1.7370526315789475e-05,
-      "loss": 1.2006,
-      "step": 1750
-    },
-    {
-      "epoch": 2.9298100743187447,
-      "grad_norm": 8.873353958129883,
-      "learning_rate": 1.731789473684211e-05,
-      "loss": 1.167,
-      "step": 1775
-    },
-    {
-      "epoch": 2.9710982658959537,
-      "grad_norm": 8.414481163024902,
-      "learning_rate": 1.726526315789474e-05,
-      "loss": 1.162,
-      "step": 1800
-    },
-    {
-      "epoch": 3.0115606936416186,
-      "grad_norm": 8.181883811950684,
-      "learning_rate": 1.721263157894737e-05,
-      "loss": 1.1539,
-      "step": 1825
-    },
-    {
-      "epoch": 3.0528488852188276,
-      "grad_norm": 8.511463165283203,
-      "learning_rate": 1.7160000000000002e-05,
-      "loss": 1.0424,
-      "step": 1850
-    },
-    {
-      "epoch": 3.094137076796036,
-      "grad_norm": 8.532715797424316,
-      "learning_rate": 1.710736842105263e-05,
-      "loss": 1.0293,
-      "step": 1875
-    },
-    {
-      "epoch": 3.135425268373245,
-      "grad_norm": 7.531360149383545,
-      "learning_rate": 1.7054736842105265e-05,
-      "loss": 0.997,
-      "step": 1900
-    },
-    {
-      "epoch": 3.176713459950454,
-      "grad_norm": 8.686123847961426,
-      "learning_rate": 1.7002105263157895e-05,
-      "loss": 1.0264,
-      "step": 1925
-    },
-    {
-      "epoch": 3.218001651527663,
-      "grad_norm": 8.670825958251953,
-      "learning_rate": 1.6949473684210528e-05,
-      "loss": 1.0235,
-      "step": 1950
-    },
-    {
-      "epoch": 3.259289843104872,
-      "grad_norm": 9.212960243225098,
-      "learning_rate": 1.689684210526316e-05,
-      "loss": 1.0109,
-      "step": 1975
-    },
-    {
-      "epoch": 3.300578034682081,
-      "grad_norm": 7.840266704559326,
-      "learning_rate": 1.684421052631579e-05,
-      "loss": 0.9955,
-      "step": 2000
-    },
-    {
-      "epoch": 3.300578034682081,
-      "eval_loss": 1.2765947580337524,
-      "eval_runtime": 710.8083,
-      "eval_samples_per_second": 14.886,
-      "eval_steps_per_second": 1.241,
-      "eval_wer": Infinity,
-      "step": 2000
-    },
-    {
-      "epoch": 3.34186622625929,
-      "grad_norm": 7.762492656707764,
-      "learning_rate": 1.6791578947368425e-05,
-      "loss": 0.9837,
-      "step": 2025
-    },
-    {
-      "epoch": 3.383154417836499,
-      "grad_norm": 8.323661804199219,
-      "learning_rate": 1.6738947368421055e-05,
-      "loss": 1.0037,
-      "step": 2050
-    },
-    {
-      "epoch": 3.424442609413708,
-      "grad_norm": 9.538601875305176,
-      "learning_rate": 1.6686315789473685e-05,
-      "loss": 1.0539,
-      "step": 2075
-    },
-    {
-      "epoch": 3.4657308009909165,
-      "grad_norm": 7.49852180480957,
-      "learning_rate": 1.6633684210526318e-05,
-      "loss": 0.9676,
-      "step": 2100
-    },
-    {
-      "epoch": 3.5070189925681254,
-      "grad_norm": 8.388888359069824,
-      "learning_rate": 1.6581052631578948e-05,
-      "loss": 1.0012,
-      "step": 2125
-    },
-    {
-      "epoch": 3.5483071841453344,
-      "grad_norm": 8.11563491821289,
-      "learning_rate": 1.652842105263158e-05,
-      "loss": 1.0151,
-      "step": 2150
-    },
-    {
-      "epoch": 3.5895953757225434,
-      "grad_norm": 8.094681739807129,
-      "learning_rate": 1.647578947368421e-05,
-      "loss": 0.9746,
-      "step": 2175
-    },
-    {
-      "epoch": 3.6308835672997524,
-      "grad_norm": 7.915094375610352,
-      "learning_rate": 1.6423157894736845e-05,
-      "loss": 1.0166,
-      "step": 2200
-    },
-    {
-      "epoch": 3.6721717588769613,
-      "grad_norm": 7.892699718475342,
-      "learning_rate": 1.6370526315789475e-05,
-      "loss": 0.9443,
-      "step": 2225
-    },
-    {
-      "epoch": 3.71345995045417,
-      "grad_norm": 8.693673133850098,
-      "learning_rate": 1.6317894736842105e-05,
-      "loss": 0.9299,
-      "step": 2250
-    },
-    {
-      "epoch": 3.754748142031379,
-      "grad_norm": 7.691691875457764,
-      "learning_rate": 1.6265263157894738e-05,
-      "loss": 0.9333,
-      "step": 2275
-    },
-    {
-      "epoch": 3.796036333608588,
-      "grad_norm": 8.604223251342773,
-      "learning_rate": 1.6212631578947368e-05,
-      "loss": 0.9782,
-      "step": 2300
-    },
-    {
-      "epoch": 3.837324525185797,
-      "grad_norm": 8.439273834228516,
-      "learning_rate": 1.616e-05,
-      "loss": 0.9229,
-      "step": 2325
-    },
-    {
-      "epoch": 3.878612716763006,
-      "grad_norm": 8.69174575805664,
-      "learning_rate": 1.6107368421052634e-05,
-      "loss": 0.9848,
-      "step": 2350
-    },
-    {
-      "epoch": 3.9199009083402148,
-      "grad_norm": 8.314153671264648,
-      "learning_rate": 1.6054736842105264e-05,
-      "loss": 0.9285,
-      "step": 2375
-    },
-    {
-      "epoch": 3.9611890999174237,
-      "grad_norm": 8.562870979309082,
-      "learning_rate": 1.6002105263157898e-05,
-      "loss": 0.9432,
-      "step": 2400
-    },
-    {
-      "epoch": 4.001651527663088,
-      "grad_norm": 7.700297832489014,
-      "learning_rate": 1.5949473684210528e-05,
-      "loss": 0.9091,
-      "step": 2425
-    },
-    {
-      "epoch": 4.042939719240297,
-      "grad_norm": 8.673540115356445,
-      "learning_rate": 1.589684210526316e-05,
-      "loss": 0.8088,
-      "step": 2450
-    },
-    {
-      "epoch": 4.084227910817506,
-      "grad_norm": 7.37542724609375,
-      "learning_rate": 1.584421052631579e-05,
-      "loss": 0.7986,
-      "step": 2475
-    },
-    {
-      "epoch": 4.125516102394715,
-      "grad_norm": 6.779449462890625,
-      "learning_rate": 1.579157894736842e-05,
-      "loss": 0.8114,
-      "step": 2500
-    },
-    {
-      "epoch": 4.166804293971924,
-      "grad_norm": 7.313191890716553,
-      "learning_rate": 1.5738947368421054e-05,
-      "loss": 0.8089,
-      "step": 2525
-    },
-    {
-      "epoch": 4.208092485549133,
-      "grad_norm": 6.947272777557373,
-      "learning_rate": 1.5686315789473684e-05,
-      "loss": 0.778,
-      "step": 2550
-    },
-    {
-      "epoch": 4.249380677126342,
-      "grad_norm": 8.322178840637207,
-      "learning_rate": 1.5633684210526318e-05,
-      "loss": 0.7897,
-      "step": 2575
-    },
-    {
-      "epoch": 4.290668868703551,
-      "grad_norm": 8.319843292236328,
-      "learning_rate": 1.5581052631578947e-05,
-      "loss": 0.8583,
-      "step": 2600
-    },
-    {
-      "epoch": 4.33195706028076,
-      "grad_norm": 7.782474994659424,
-      "learning_rate": 1.552842105263158e-05,
-      "loss": 0.8077,
-      "step": 2625
-    },
-    {
-      "epoch": 4.373245251857969,
-      "grad_norm": 8.119930267333984,
-      "learning_rate": 1.547578947368421e-05,
-      "loss": 0.807,
-      "step": 2650
-    },
-    {
-      "epoch": 4.414533443435178,
-      "grad_norm": 10.038588523864746,
-      "learning_rate": 1.5423157894736844e-05,
-      "loss": 0.8248,
-      "step": 2675
-    },
-    {
-      "epoch": 4.455821635012386,
-      "grad_norm": 7.1363115310668945,
-      "learning_rate": 1.5370526315789477e-05,
-      "loss": 0.8092,
-      "step": 2700
-    },
-    {
-      "epoch": 4.497109826589595,
-      "grad_norm": 8.094152450561523,
-      "learning_rate": 1.5317894736842107e-05,
-      "loss": 0.8386,
-      "step": 2725
-    },
-    {
-      "epoch": 4.538398018166804,
-      "grad_norm": 8.467181205749512,
-      "learning_rate": 1.5265263157894737e-05,
-      "loss": 0.802,
-      "step": 2750
-    },
-    {
-      "epoch": 4.579686209744013,
-      "grad_norm": 9.26181697845459,
-      "learning_rate": 1.521263157894737e-05,
-      "loss": 0.8057,
-      "step": 2775
-    },
-    {
-      "epoch": 4.620974401321222,
-      "grad_norm": 8.754571914672852,
-      "learning_rate": 1.516e-05,
-      "loss": 0.7905,
-      "step": 2800
-    },
-    {
-      "epoch": 4.662262592898431,
-      "grad_norm": 7.060452461242676,
-      "learning_rate": 1.5107368421052632e-05,
-      "loss": 0.7644,
-      "step": 2825
-    },
-    {
-      "epoch": 4.70355078447564,
-      "grad_norm": 8.613765716552734,
-      "learning_rate": 1.5054736842105264e-05,
-      "loss": 0.7676,
-      "step": 2850
-    },
-    {
-      "epoch": 4.744838976052849,
-      "grad_norm": 7.17141056060791,
-      "learning_rate": 1.5002105263157895e-05,
-      "loss": 0.7648,
-      "step": 2875
-    },
-    {
-      "epoch": 4.786127167630058,
-      "grad_norm": 8.803303718566895,
-      "learning_rate": 1.4949473684210527e-05,
-      "loss": 0.8048,
-      "step": 2900
-    },
-    {
-      "epoch": 4.827415359207267,
-      "grad_norm": 7.4490203857421875,
-      "learning_rate": 1.4896842105263159e-05,
-      "loss": 0.78,
-      "step": 2925
-    },
-    {
-      "epoch": 4.868703550784476,
-      "grad_norm": 8.119257926940918,
-      "learning_rate": 1.484421052631579e-05,
-      "loss": 0.7647,
-      "step": 2950
-    },
-    {
-      "epoch": 4.909991742361685,
-      "grad_norm": 7.520335674285889,
-      "learning_rate": 1.4791578947368422e-05,
-      "loss": 0.772,
-      "step": 2975
-    },
-    {
-      "epoch": 4.951279933938894,
-      "grad_norm": 7.8852219581604,
-      "learning_rate": 1.4738947368421055e-05,
-      "loss": 0.7584,
-      "step": 3000
-    },
-    {
-      "epoch": 4.951279933938894,
-      "eval_loss": 1.1080505847930908,
-      "eval_runtime": 699.2603,
-      "eval_samples_per_second": 15.132,
-      "eval_steps_per_second": 1.261,
-      "eval_wer": Infinity,
-      "step": 3000
-    },
-    {
-      "epoch": 4.992568125516103,
-      "grad_norm": 8.253033638000488,
-      "learning_rate": 1.4686315789473687e-05,
-      "loss": 0.7497,
-      "step": 3025
-    },
-    {
-      "epoch": 5.033030553261767,
-      "grad_norm": 7.215710163116455,
-      "learning_rate": 1.4633684210526317e-05,
-      "loss": 0.6874,
-      "step": 3050
-    },
-    {
-      "epoch": 5.074318744838976,
-      "grad_norm": 7.491812229156494,
-      "learning_rate": 1.4581052631578949e-05,
-      "loss": 0.6607,
-      "step": 3075
-    },
-    {
-      "epoch": 5.115606936416185,
-      "grad_norm": 7.644138813018799,
-      "learning_rate": 1.452842105263158e-05,
-      "loss": 0.6834,
-      "step": 3100
-    },
-    {
-      "epoch": 5.156895127993394,
-      "grad_norm": 7.1401686668396,
-      "learning_rate": 1.4475789473684212e-05,
-      "loss": 0.6667,
-      "step": 3125
-    },
-    {
-      "epoch": 5.198183319570603,
-      "grad_norm": 7.159341335296631,
-      "learning_rate": 1.4423157894736843e-05,
-      "loss": 0.6706,
-      "step": 3150
-    },
-    {
-      "epoch": 5.239471511147812,
-      "grad_norm": 6.769131660461426,
-      "learning_rate": 1.4370526315789475e-05,
-      "loss": 0.6508,
-      "step": 3175
-    },
-    {
-      "epoch": 5.280759702725021,
-      "grad_norm": 7.549663543701172,
-      "learning_rate": 1.4317894736842107e-05,
-      "loss": 0.6674,
-      "step": 3200
-    },
-    {
-      "epoch": 5.32204789430223,
-      "grad_norm": 7.3631720542907715,
-      "learning_rate": 1.4265263157894738e-05,
-      "loss": 0.6401,
-      "step": 3225
-    },
-    {
-      "epoch": 5.363336085879438,
-      "grad_norm": 7.147627353668213,
-      "learning_rate": 1.4212631578947368e-05,
-      "loss": 0.6805,
-      "step": 3250
-    },
-    {
-      "epoch": 5.404624277456647,
-      "grad_norm": 7.807270050048828,
-      "learning_rate": 1.416e-05,
-      "loss": 0.6333,
-      "step": 3275
-    },
-    {
-      "epoch": 5.445912469033856,
-      "grad_norm": 7.820711612701416,
-      "learning_rate": 1.4107368421052632e-05,
-      "loss": 0.7001,
-      "step": 3300
-    },
-    {
-      "epoch": 5.487200660611065,
-      "grad_norm": 8.526030540466309,
-      "learning_rate": 1.4054736842105263e-05,
-      "loss": 0.6728,
-      "step": 3325
-    },
-    {
-      "epoch": 5.528488852188274,
-      "grad_norm": 6.857508659362793,
-      "learning_rate": 1.4002105263157897e-05,
-      "loss": 0.6425,
-      "step": 3350
-    },
-    {
-      "epoch": 5.569777043765483,
-      "grad_norm": 6.723370552062988,
-      "learning_rate": 1.3949473684210528e-05,
-      "loss": 0.6616,
-      "step": 3375
-    },
-    {
-      "epoch": 5.611065235342692,
-      "grad_norm": 7.373733043670654,
-      "learning_rate": 1.389684210526316e-05,
-      "loss": 0.6609,
-      "step": 3400
-    },
-    {
-      "epoch": 5.652353426919901,
-      "grad_norm": 7.192294120788574,
-      "learning_rate": 1.3844210526315791e-05,
-      "loss": 0.6413,
-      "step": 3425
-    },
-    {
-      "epoch": 5.69364161849711,
-      "grad_norm": 6.876883029937744,
-      "learning_rate": 1.3791578947368423e-05,
-      "loss": 0.648,
-      "step": 3450
-    },
-    {
-      "epoch": 5.734929810074319,
-      "grad_norm": 7.807666301727295,
-      "learning_rate": 1.3738947368421055e-05,
-      "loss": 0.6643,
-      "step": 3475
-    },
-    {
-      "epoch": 5.776218001651528,
-      "grad_norm": 7.850846767425537,
-      "learning_rate": 1.3686315789473685e-05,
-      "loss": 0.6469,
-      "step": 3500
-    },
-    {
-      "epoch": 5.817506193228737,
-      "grad_norm": 7.765756607055664,
-      "learning_rate": 1.3633684210526316e-05,
-      "loss": 0.6642,
-      "step": 3525
-    },
-    {
-      "epoch": 5.858794384805946,
-      "grad_norm": 7.8592915534973145,
-      "learning_rate": 1.3581052631578948e-05,
-      "loss": 0.6531,
-      "step": 3550
-    },
-    {
-      "epoch": 5.900082576383154,
-      "grad_norm": 7.569016456604004,
-      "learning_rate": 1.352842105263158e-05,
-      "loss": 0.646,
-      "step": 3575
-    },
-    {
-      "epoch": 5.941370767960363,
-      "grad_norm": 6.537998199462891,
-      "learning_rate": 1.3475789473684211e-05,
-      "loss": 0.609,
-      "step": 3600
-    },
-    {
-      "epoch": 5.982658959537572,
-      "grad_norm": 7.915237903594971,
-      "learning_rate": 1.3423157894736843e-05,
-      "loss": 0.6304,
-      "step": 3625
-    },
-    {
-      "epoch": 6.023121387283237,
-      "grad_norm": 7.4267683029174805,
-      "learning_rate": 1.3370526315789475e-05,
-      "loss": 0.5556,
-      "step": 3650
-    },
-    {
-      "epoch": 6.064409578860446,
-      "grad_norm": 6.749971389770508,
-      "learning_rate": 1.3317894736842108e-05,
-      "loss": 0.5266,
-      "step": 3675
-    },
-    {
-      "epoch": 6.105697770437655,
-      "grad_norm": 6.487602233886719,
-      "learning_rate": 1.326526315789474e-05,
-      "loss": 0.5717,
-      "step": 3700
-    },
-    {
-      "epoch": 6.146985962014864,
-      "grad_norm": 6.367309093475342,
-      "learning_rate": 1.321263157894737e-05,
-      "loss": 0.5661,
-      "step": 3725
-    },
-    {
-      "epoch": 6.188274153592072,
-      "grad_norm": 6.503593444824219,
-      "learning_rate": 1.3160000000000001e-05,
-      "loss": 0.552,
-      "step": 3750
-    },
-    {
-      "epoch": 6.229562345169281,
-      "grad_norm": 6.83872127532959,
-      "learning_rate": 1.3107368421052633e-05,
-      "loss": 0.57,
-      "step": 3775
-    },
-    {
-      "epoch": 6.27085053674649,
-      "grad_norm": 6.761080741882324,
-      "learning_rate": 1.3054736842105264e-05,
-      "loss": 0.5332,
-      "step": 3800
-    },
-    {
-      "epoch": 6.312138728323699,
-      "grad_norm": 7.023492336273193,
-      "learning_rate": 1.3002105263157896e-05,
-      "loss": 0.5588,
-      "step": 3825
-    },
-    {
-      "epoch": 6.353426919900908,
-      "grad_norm": 7.079690933227539,
-      "learning_rate": 1.2949473684210528e-05,
-      "loss": 0.5471,
-      "step": 3850
-    },
-    {
-      "epoch": 6.394715111478117,
-      "grad_norm": 7.1418633460998535,
-      "learning_rate": 1.289684210526316e-05,
-      "loss": 0.5471,
-      "step": 3875
-    },
-    {
-      "epoch": 6.436003303055326,
-      "grad_norm": 6.428386211395264,
-      "learning_rate": 1.2844210526315791e-05,
-      "loss": 0.5519,
-      "step": 3900
-    },
-    {
-      "epoch": 6.477291494632535,
-      "grad_norm": 6.753655910491943,
-      "learning_rate": 1.279157894736842e-05,
-      "loss": 0.5745,
-      "step": 3925
-    },
-    {
-      "epoch": 6.518579686209744,
-      "grad_norm": 6.977758407592773,
-      "learning_rate": 1.2738947368421052e-05,
-      "loss": 0.5657,
-      "step": 3950
-    },
-    {
-      "epoch": 6.559867877786953,
-      "grad_norm": 6.839511394500732,
-      "learning_rate": 1.2686315789473684e-05,
-      "loss": 0.5432,
-      "step": 3975
-    },
-    {
-      "epoch": 6.601156069364162,
-      "grad_norm": 7.744399547576904,
-      "learning_rate": 1.2633684210526316e-05,
-      "loss": 0.5473,
-      "step": 4000
-    },
-    {
-      "epoch": 6.601156069364162,
-      "eval_loss": 1.056926965713501,
-      "eval_runtime": 696.4132,
-      "eval_samples_per_second": 15.194,
-      "eval_steps_per_second": 1.266,
-      "eval_wer": Infinity,
-      "step": 4000
-    },
-    {
-      "epoch": 6.642444260941371,
-      "grad_norm": 7.011012077331543,
-      "learning_rate": 1.2581052631578949e-05,
-      "loss": 0.5369,
-      "step": 4025
-    },
-    {
-      "epoch": 6.68373245251858,
-      "grad_norm": 8.44078540802002,
-      "learning_rate": 1.252842105263158e-05,
-      "loss": 0.5776,
-      "step": 4050
-    },
-    {
-      "epoch": 6.725020644095789,
-      "grad_norm": 7.306187629699707,
-      "learning_rate": 1.2475789473684212e-05,
-      "loss": 0.5903,
-      "step": 4075
-    },
-    {
-      "epoch": 6.766308835672998,
-      "grad_norm": 6.952062606811523,
-      "learning_rate": 1.2423157894736844e-05,
-      "loss": 0.55,
-      "step": 4100
-    },
-    {
-      "epoch": 6.807597027250207,
-      "grad_norm": 6.906418323516846,
-      "learning_rate": 1.2370526315789476e-05,
-      "loss": 0.528,
-      "step": 4125
-    },
-    {
-      "epoch": 6.848885218827416,
-      "grad_norm": 7.448940277099609,
-      "learning_rate": 1.2317894736842107e-05,
-      "loss": 0.5505,
-      "step": 4150
-    },
-    {
-      "epoch": 6.890173410404624,
-      "grad_norm": 7.480178356170654,
-      "learning_rate": 1.2265263157894737e-05,
-      "loss": 0.5409,
-      "step": 4175
-    },
-    {
-      "epoch": 6.931461601981833,
-      "grad_norm": 7.500922679901123,
-      "learning_rate": 1.2212631578947369e-05,
-      "loss": 0.5136,
-      "step": 4200
-    },
-    {
-      "epoch": 6.972749793559042,
-      "grad_norm": 6.731492042541504,
-      "learning_rate": 1.216e-05,
-      "loss": 0.534,
-      "step": 4225
-    },
-    {
-      "epoch": 7.013212221304707,
-      "grad_norm": 5.806182861328125,
-      "learning_rate": 1.2107368421052632e-05,
-      "loss": 0.4876,
-      "step": 4250
-    },
-    {
-      "epoch": 7.054500412881916,
-      "grad_norm": 6.7710089683532715,
-      "learning_rate": 1.2054736842105264e-05,
-      "loss": 0.4452,
-      "step": 4275
-    },
-    {
-      "epoch": 7.095788604459124,
-      "grad_norm": 7.5482563972473145,
-      "learning_rate": 1.2002105263157895e-05,
-      "loss": 0.4388,
-      "step": 4300
-    },
-    {
-      "epoch": 7.137076796036333,
-      "grad_norm": 5.3507537841796875,
-      "learning_rate": 1.1949473684210527e-05,
-      "loss": 0.4572,
-      "step": 4325
-    },
-    {
-      "epoch": 7.178364987613542,
-      "grad_norm": 6.261397838592529,
-      "learning_rate": 1.189684210526316e-05,
-      "loss": 0.46,
-      "step": 4350
-    },
-    {
-      "epoch": 7.219653179190751,
-      "grad_norm": 7.379156589508057,
-      "learning_rate": 1.1844210526315792e-05,
-      "loss": 0.4595,
-      "step": 4375
-    },
-    {
-      "epoch": 7.26094137076796,
-      "grad_norm": 6.441304683685303,
-      "learning_rate": 1.1791578947368424e-05,
-      "loss": 0.4623,
-      "step": 4400
-    },
-    {
-      "epoch": 7.302229562345169,
-      "grad_norm": 6.100840091705322,
-      "learning_rate": 1.1738947368421054e-05,
-      "loss": 0.4686,
-      "step": 4425
-    },
-    {
-      "epoch": 7.343517753922378,
-      "grad_norm": 6.594038486480713,
-      "learning_rate": 1.1686315789473685e-05,
-      "loss": 0.4638,
-      "step": 4450
-    },
-    {
-      "epoch": 7.384805945499587,
-      "grad_norm": 6.223761558532715,
-      "learning_rate": 1.1633684210526317e-05,
-      "loss": 0.4687,
-      "step": 4475
-    },
-    {
-      "epoch": 7.426094137076796,
-      "grad_norm": 5.7157063484191895,
-      "learning_rate": 1.1581052631578948e-05,
-      "loss": 0.4475,
-      "step": 4500
-    },
-    {
-      "epoch": 7.467382328654005,
-      "grad_norm": 7.293723106384277,
-      "learning_rate": 1.152842105263158e-05,
-      "loss": 0.4725,
-      "step": 4525
-    },
-    {
-      "epoch": 7.508670520231214,
-      "grad_norm": 6.591246604919434,
-      "learning_rate": 1.1475789473684212e-05,
-      "loss": 0.4783,
-      "step": 4550
-    },
-    {
-      "epoch": 7.549958711808423,
-      "grad_norm": 7.094688415527344,
-      "learning_rate": 1.1423157894736843e-05,
-      "loss": 0.4521,
-      "step": 4575
-    },
-    {
-      "epoch": 7.591246903385632,
-      "grad_norm": 7.4666595458984375,
-      "learning_rate": 1.1370526315789473e-05,
-      "loss": 0.4679,
-      "step": 4600
-    },
-    {
-      "epoch": 7.632535094962841,
-      "grad_norm": 6.891570568084717,
-      "learning_rate": 1.1317894736842105e-05,
-      "loss": 0.469,
-      "step": 4625
-    },
-    {
-      "epoch": 7.673823286540049,
-      "grad_norm": 7.736660003662109,
-      "learning_rate": 1.1265263157894737e-05,
-      "loss": 0.4874,
-      "step": 4650
-    },
-    {
-      "epoch": 7.715111478117258,
-      "grad_norm": 7.7448248863220215,
-      "learning_rate": 1.1212631578947368e-05,
-      "loss": 0.4516,
-      "step": 4675
-    },
-    {
-      "epoch": 7.756399669694467,
-      "grad_norm": 6.311560153961182,
-      "learning_rate": 1.1160000000000002e-05,
-      "loss": 0.4642,
-      "step": 4700
-    },
-    {
-      "epoch": 7.797687861271676,
-      "grad_norm": 7.201409816741943,
-      "learning_rate": 1.1107368421052633e-05,
-      "loss": 0.4545,
-      "step": 4725
-    },
-    {
-      "epoch": 7.838976052848885,
-      "grad_norm": 6.123779773712158,
-      "learning_rate": 1.1054736842105265e-05,
-      "loss": 0.4287,
-      "step": 4750
-    },
-    {
-      "epoch": 7.880264244426094,
-      "grad_norm": 7.237743377685547,
-      "learning_rate": 1.1002105263157896e-05,
-      "loss": 0.4993,
-      "step": 4775
-    },
-    {
-      "epoch": 7.921552436003303,
-      "grad_norm": 6.445917129516602,
-      "learning_rate": 1.0949473684210528e-05,
-      "loss": 0.4843,
-      "step": 4800
-    },
-    {
-      "epoch": 7.962840627580512,
-      "grad_norm": 6.234178066253662,
-      "learning_rate": 1.089684210526316e-05,
-      "loss": 0.4657,
-      "step": 4825
-    },
-    {
-      "epoch": 8.003303055326176,
-      "grad_norm": 7.247951507568359,
-      "learning_rate": 1.084421052631579e-05,
-      "loss": 0.4923,
-      "step": 4850
-    },
-    {
-      "epoch": 8.044591246903385,
-      "grad_norm": 5.9410223960876465,
-      "learning_rate": 1.0791578947368421e-05,
-      "loss": 0.3872,
-      "step": 4875
-    },
-    {
-      "epoch": 8.085879438480594,
-      "grad_norm": 6.340057849884033,
-      "learning_rate": 1.0738947368421053e-05,
-      "loss": 0.3966,
-      "step": 4900
-    },
-    {
-      "epoch": 8.127167630057803,
-      "grad_norm": 6.049722194671631,
-      "learning_rate": 1.0686315789473685e-05,
-      "loss": 0.3706,
-      "step": 4925
-    },
-    {
-      "epoch": 8.168455821635012,
-      "grad_norm": 5.964404582977295,
-      "learning_rate": 1.0633684210526316e-05,
-      "loss": 0.3854,
-      "step": 4950
-    },
-    {
-      "epoch": 8.209744013212221,
-      "grad_norm": 6.412349224090576,
-      "learning_rate": 1.0581052631578948e-05,
-      "loss": 0.3923,
-      "step": 4975
-    },
-    {
-      "epoch": 8.25103220478943,
-      "grad_norm": 6.039672374725342,
-      "learning_rate": 1.052842105263158e-05,
-      "loss": 0.4191,
-      "step": 5000
-    },
-    {
-      "epoch": 8.25103220478943,
-      "eval_loss": 1.0567598342895508,
-      "eval_runtime": 694.1236,
-      "eval_samples_per_second": 15.244,
-      "eval_steps_per_second": 1.271,
-      "eval_wer": Infinity,
-      "step": 5000
-    },
-    {
-      "epoch": 8.29232039636664,
-      "grad_norm": 6.2119011878967285,
-      "learning_rate": 1.047578947368421e-05,
-      "loss": 0.3807,
-      "step": 5025
-    },
-    {
-      "epoch": 8.333608587943848,
-      "grad_norm": 6.253497123718262,
-      "learning_rate": 1.0423157894736844e-05,
-      "loss": 0.3827,
-      "step": 5050
-    },
-    {
-      "epoch": 8.374896779521057,
-      "grad_norm": 6.059364318847656,
-      "learning_rate": 1.0370526315789476e-05,
-      "loss": 0.389,
-      "step": 5075
-    },
-    {
-      "epoch": 8.416184971098266,
-      "grad_norm": 5.26777982711792,
-      "learning_rate": 1.0317894736842106e-05,
-      "loss": 0.3998,
-      "step": 5100
-    },
-    {
-      "epoch": 8.457473162675475,
-      "grad_norm": 5.529355525970459,
-      "learning_rate": 1.0265263157894738e-05,
-      "loss": 0.3954,
-      "step": 5125
-    },
-    {
-      "epoch": 8.498761354252684,
-      "grad_norm": 6.751101016998291,
-      "learning_rate": 1.021263157894737e-05,
-      "loss": 0.3964,
-      "step": 5150
-    },
-    {
-      "epoch": 8.540049545829893,
-      "grad_norm": 5.405825138092041,
-      "learning_rate": 1.0160000000000001e-05,
-      "loss": 0.389,
-      "step": 5175
-    },
-    {
-      "epoch": 8.581337737407102,
-      "grad_norm": 7.287736415863037,
-      "learning_rate": 1.0107368421052633e-05,
-      "loss": 0.4196,
-      "step": 5200
-    },
-    {
-      "epoch": 8.622625928984311,
-      "grad_norm": 6.384584903717041,
-      "learning_rate": 1.0054736842105264e-05,
-      "loss": 0.4133,
-      "step": 5225
-    },
-    {
-      "epoch": 8.66391412056152,
-      "grad_norm": 6.451061725616455,
-      "learning_rate": 1.0002105263157896e-05,
-      "loss": 0.3837,
-      "step": 5250
-    },
-    {
-      "epoch": 8.705202312138729,
-      "grad_norm": 5.920280933380127,
-      "learning_rate": 9.949473684210526e-06,
-      "loss": 0.4004,
-      "step": 5275
-    },
-    {
-      "epoch": 8.746490503715938,
-      "grad_norm": 5.891169548034668,
-      "learning_rate": 9.89684210526316e-06,
-      "loss": 0.3932,
-      "step": 5300
-    },
-    {
-      "epoch": 8.787778695293147,
-      "grad_norm": 6.570876598358154,
-      "learning_rate": 9.84421052631579e-06,
-      "loss": 0.4032,
-      "step": 5325
-    },
-    {
-      "epoch": 8.829066886870356,
-      "grad_norm": 7.074426174163818,
-      "learning_rate": 9.791578947368422e-06,
-      "loss": 0.3951,
-      "step": 5350
-    },
-    {
-      "epoch": 8.870355078447563,
-      "grad_norm": 6.7733564376831055,
-      "learning_rate": 9.738947368421054e-06,
-      "loss": 0.3992,
-      "step": 5375
-    },
-    {
-      "epoch": 8.911643270024772,
-      "grad_norm": 6.8463969230651855,
-      "learning_rate": 9.686315789473684e-06,
-      "loss": 0.4004,
-      "step": 5400
-    },
-    {
-      "epoch": 8.952931461601981,
-      "grad_norm": 6.444080352783203,
-      "learning_rate": 9.633684210526316e-06,
-      "loss": 0.3844,
-      "step": 5425
-    },
-    {
-      "epoch": 8.99421965317919,
-      "grad_norm": 6.648684024810791,
-      "learning_rate": 9.581052631578947e-06,
-      "loss": 0.3962,
-      "step": 5450
-    },
-    {
-      "epoch": 9.034682080924856,
-      "grad_norm": 5.462916851043701,
-      "learning_rate": 9.52842105263158e-06,
-      "loss": 0.3186,
-      "step": 5475
-    },
-    {
-      "epoch": 9.075970272502065,
-      "grad_norm": 4.91270112991333,
-      "learning_rate": 9.475789473684212e-06,
-      "loss": 0.3091,
-      "step": 5500
-    },
-    {
-      "epoch": 9.117258464079274,
-      "grad_norm": 6.3385467529296875,
-      "learning_rate": 9.423157894736842e-06,
-      "loss": 0.3088,
-      "step": 5525
-    },
-    {
-      "epoch": 9.158546655656481,
-      "grad_norm": 6.092479228973389,
-      "learning_rate": 9.370526315789474e-06,
-      "loss": 0.3366,
-      "step": 5550
-    },
-    {
-      "epoch": 9.19983484723369,
-      "grad_norm": 6.2821364402771,
-      "learning_rate": 9.317894736842105e-06,
-      "loss": 0.328,
-      "step": 5575
-    },
-    {
-      "epoch": 9.2411230388109,
-      "grad_norm": 6.607597827911377,
-      "learning_rate": 9.265263157894737e-06,
-      "loss": 0.3237,
-      "step": 5600
-    },
-    {
-      "epoch": 9.282411230388108,
-      "grad_norm": 6.044376850128174,
-      "learning_rate": 9.21263157894737e-06,
-      "loss": 0.3405,
-      "step": 5625
-    },
-    {
-      "epoch": 9.323699421965317,
-      "grad_norm": 5.687169551849365,
-      "learning_rate": 9.16e-06,
-      "loss": 0.3433,
-      "step": 5650
-    },
-    {
-      "epoch": 9.364987613542526,
-      "grad_norm": 6.336760997772217,
-      "learning_rate": 9.107368421052632e-06,
-      "loss": 0.3681,
-      "step": 5675
-    },
-    {
-      "epoch": 9.406275805119735,
-      "grad_norm": 5.349738597869873,
-      "learning_rate": 9.054736842105264e-06,
-      "loss": 0.3414,
-      "step": 5700
-    },
-    {
-      "epoch": 9.447563996696944,
-      "grad_norm": 5.843466758728027,
-      "learning_rate": 9.002105263157895e-06,
-      "loss": 0.3494,
-      "step": 5725
-    },
-    {
-      "epoch": 9.488852188274153,
-      "grad_norm": 5.646897315979004,
-      "learning_rate": 8.949473684210527e-06,
-      "loss": 0.3491,
-      "step": 5750
-    },
-    {
-      "epoch": 9.530140379851362,
-      "grad_norm": 6.2915472984313965,
-      "learning_rate": 8.896842105263159e-06,
-      "loss": 0.3627,
-      "step": 5775
-    },
-    {
-      "epoch": 9.571428571428571,
-      "grad_norm": 5.424402236938477,
-      "learning_rate": 8.84421052631579e-06,
-      "loss": 0.3203,
-      "step": 5800
-    },
-    {
-      "epoch": 9.61271676300578,
-      "grad_norm": 6.5985260009765625,
-      "learning_rate": 8.791578947368422e-06,
-      "loss": 0.3449,
-      "step": 5825
-    },
-    {
-      "epoch": 9.65400495458299,
-      "grad_norm": 5.538530349731445,
-      "learning_rate": 8.738947368421053e-06,
-      "loss": 0.3407,
-      "step": 5850
-    },
-    {
-      "epoch": 9.695293146160198,
-      "grad_norm": 5.327653408050537,
-      "learning_rate": 8.686315789473685e-06,
-      "loss": 0.3235,
-      "step": 5875
-    },
-    {
-      "epoch": 9.736581337737407,
-      "grad_norm": 6.340036392211914,
-      "learning_rate": 8.633684210526317e-06,
-      "loss": 0.3595,
-      "step": 5900
-    },
-    {
-      "epoch": 9.777869529314616,
-      "grad_norm": 5.395780086517334,
-      "learning_rate": 8.581052631578948e-06,
-      "loss": 0.3368,
-      "step": 5925
-    },
-    {
-      "epoch": 9.819157720891825,
-      "grad_norm": 5.803914546966553,
-      "learning_rate": 8.528421052631578e-06,
-      "loss": 0.3497,
-      "step": 5950
-    },
-    {
-      "epoch": 9.860445912469034,
-      "grad_norm": 6.23885440826416,
-      "learning_rate": 8.475789473684212e-06,
-      "loss": 0.3504,
-      "step": 5975
-    },
-    {
-      "epoch": 9.901734104046243,
-      "grad_norm": 6.402750015258789,
-      "learning_rate": 8.425263157894737e-06,
-      "loss": 0.3167,
-      "step": 6000
-    },
-    {
-      "epoch": 9.901734104046243,
-      "eval_loss": 1.060942530632019,
-      "eval_runtime": 695.1498,
-      "eval_samples_per_second": 15.221,
-      "eval_steps_per_second": 1.269,
-      "eval_wer": Infinity,
-      "step": 6000
-    },
-    {
-      "epoch": 9.943022295623452,
-      "grad_norm": 5.562705993652344,
-      "learning_rate": 8.372631578947368e-06,
-      "loss": 0.3252,
-      "step": 6025
-    },
-    {
-      "epoch": 9.984310487200661,
-      "grad_norm": 7.461434364318848,
-      "learning_rate": 8.32e-06,
-      "loss": 0.344,
-      "step": 6050
-    },
-    {
-      "epoch": 10.024772914946325,
-      "grad_norm": 5.992921352386475,
-      "learning_rate": 8.267368421052632e-06,
-      "loss": 0.2875,
-      "step": 6075
-    },
-    {
-      "epoch": 10.066061106523534,
-      "grad_norm": 4.986021995544434,
-      "learning_rate": 8.214736842105265e-06,
-      "loss": 0.2773,
-      "step": 6100
-    },
-    {
-      "epoch": 10.107349298100743,
-      "grad_norm": 4.8151092529296875,
-      "learning_rate": 8.162105263157895e-06,
-      "loss": 0.2886,
-      "step": 6125
-    },
-    {
-      "epoch": 10.148637489677952,
-      "grad_norm": 4.7961883544921875,
-      "learning_rate": 8.109473684210527e-06,
-      "loss": 0.3002,
-      "step": 6150
-    },
-    {
-      "epoch": 10.189925681255161,
-      "grad_norm": 5.7180399894714355,
-      "learning_rate": 8.056842105263158e-06,
-      "loss": 0.2995,
-      "step": 6175
-    },
-    {
-      "epoch": 10.23121387283237,
-      "grad_norm": 5.287256717681885,
-      "learning_rate": 8.00421052631579e-06,
-      "loss": 0.2723,
-      "step": 6200
-    },
-    {
-      "epoch": 10.27250206440958,
-      "grad_norm": 6.107348442077637,
-      "learning_rate": 7.951578947368421e-06,
-      "loss": 0.2909,
-      "step": 6225
-    },
-    {
-      "epoch": 10.313790255986788,
-      "grad_norm": 5.607096195220947,
-      "learning_rate": 7.898947368421053e-06,
-      "loss": 0.3019,
-      "step": 6250
-    },
-    {
-      "epoch": 10.355078447563997,
-      "grad_norm": 5.941197872161865,
-      "learning_rate": 7.846315789473685e-06,
-      "loss": 0.29,
-      "step": 6275
-    },
-    {
-      "epoch": 10.396366639141206,
-      "grad_norm": 5.736889362335205,
-      "learning_rate": 7.793684210526316e-06,
-      "loss": 0.3138,
-      "step": 6300
-    },
-    {
-      "epoch": 10.437654830718415,
-      "grad_norm": 5.2913126945495605,
-      "learning_rate": 7.741052631578948e-06,
-      "loss": 0.28,
-      "step": 6325
-    },
-    {
-      "epoch": 10.478943022295624,
-      "grad_norm": 5.760861396789551,
-      "learning_rate": 7.68842105263158e-06,
-      "loss": 0.3069,
-      "step": 6350
-    },
-    {
-      "epoch": 10.520231213872833,
-      "grad_norm": 4.886830806732178,
-      "learning_rate": 7.635789473684211e-06,
-      "loss": 0.2785,
-      "step": 6375
-    },
-    {
-      "epoch": 10.561519405450042,
-      "grad_norm": 5.5294389724731445,
-      "learning_rate": 7.583157894736842e-06,
-      "loss": 0.2862,
-      "step": 6400
-    },
-    {
-      "epoch": 10.602807597027251,
-      "grad_norm": 5.370204448699951,
-      "learning_rate": 7.5305263157894745e-06,
-      "loss": 0.2794,
-      "step": 6425
-    },
-    {
-      "epoch": 10.64409578860446,
-      "grad_norm": 4.194797039031982,
-      "learning_rate": 7.477894736842106e-06,
-      "loss": 0.3064,
-      "step": 6450
-    },
-    {
-      "epoch": 10.685383980181667,
-      "grad_norm": 6.214608669281006,
-      "learning_rate": 7.425263157894738e-06,
-      "loss": 0.3177,
-      "step": 6475
-    },
-    {
-      "epoch": 10.726672171758876,
-      "grad_norm": 5.77274227142334,
-      "learning_rate": 7.3726315789473694e-06,
-      "loss": 0.31,
-      "step": 6500
-    },
-    {
-      "epoch": 10.767960363336085,
-      "grad_norm": 6.4990105628967285,
-      "learning_rate": 7.32e-06,
-      "loss": 0.2716,
-      "step": 6525
-    },
-    {
-      "epoch": 10.809248554913294,
-      "grad_norm": 5.706321716308594,
-      "learning_rate": 7.267368421052632e-06,
-      "loss": 0.2983,
-      "step": 6550
-    },
-    {
-      "epoch": 10.850536746490503,
-      "grad_norm": 6.846646308898926,
-      "learning_rate": 7.2147368421052635e-06,
-      "loss": 0.2678,
-      "step": 6575
-    },
-    {
-      "epoch": 10.891824938067712,
-      "grad_norm": 5.539628982543945,
-      "learning_rate": 7.162105263157896e-06,
-      "loss": 0.2809,
-      "step": 6600
-    },
-    {
-      "epoch": 10.933113129644921,
-      "grad_norm": 5.272496700286865,
-      "learning_rate": 7.109473684210528e-06,
-      "loss": 0.2473,
-      "step": 6625
-    },
-    {
-      "epoch": 10.97440132122213,
-      "grad_norm": 5.260607719421387,
-      "learning_rate": 7.056842105263158e-06,
-      "loss": 0.2931,
-      "step": 6650
-    },
-    {
-      "epoch": 11.014863748967795,
-      "grad_norm": 5.241586685180664,
-      "learning_rate": 7.00421052631579e-06,
-      "loss": 0.2771,
-      "step": 6675
-    },
-    {
-      "epoch": 11.056151940545003,
-      "grad_norm": 4.1120781898498535,
-      "learning_rate": 6.951578947368422e-06,
-      "loss": 0.2511,
-      "step": 6700
-    },
-    {
-      "epoch": 11.097440132122212,
-      "grad_norm": 5.060145378112793,
-      "learning_rate": 6.8989473684210525e-06,
-      "loss": 0.2502,
-      "step": 6725
-    },
-    {
-      "epoch": 11.138728323699421,
-      "grad_norm": 6.010138988494873,
-      "learning_rate": 6.846315789473684e-06,
-      "loss": 0.2442,
-      "step": 6750
-    },
-    {
-      "epoch": 11.18001651527663,
-      "grad_norm": 5.282010078430176,
-      "learning_rate": 6.793684210526317e-06,
-      "loss": 0.2485,
-      "step": 6775
-    },
-    {
-      "epoch": 11.22130470685384,
-      "grad_norm": 4.690225601196289,
-      "learning_rate": 6.741052631578948e-06,
-      "loss": 0.2506,
-      "step": 6800
-    },
-    {
-      "epoch": 11.262592898431048,
-      "grad_norm": 3.9595465660095215,
-      "learning_rate": 6.68842105263158e-06,
-      "loss": 0.2325,
-      "step": 6825
-    },
-    {
-      "epoch": 11.303881090008257,
-      "grad_norm": 5.889693737030029,
-      "learning_rate": 6.635789473684211e-06,
-      "loss": 0.2527,
-      "step": 6850
-    },
-    {
-      "epoch": 11.345169281585466,
-      "grad_norm": 5.33756685256958,
-      "learning_rate": 6.583157894736842e-06,
-      "loss": 0.2488,
-      "step": 6875
-    },
-    {
-      "epoch": 11.386457473162675,
-      "grad_norm": 5.049898147583008,
-      "learning_rate": 6.530526315789474e-06,
-      "loss": 0.2355,
-      "step": 6900
-    },
-    {
-      "epoch": 11.427745664739884,
-      "grad_norm": 5.452274799346924,
-      "learning_rate": 6.477894736842106e-06,
-      "loss": 0.249,
-      "step": 6925
-    },
-    {
-      "epoch": 11.469033856317093,
-      "grad_norm": 4.464880466461182,
-      "learning_rate": 6.425263157894738e-06,
-      "loss": 0.2523,
-      "step": 6950
-    },
-    {
-      "epoch": 11.510322047894302,
-      "grad_norm": 5.923793792724609,
-      "learning_rate": 6.372631578947369e-06,
-      "loss": 0.2564,
-      "step": 6975
-    },
-    {
-      "epoch": 11.551610239471511,
-      "grad_norm": 4.821681976318359,
-      "learning_rate": 6.3200000000000005e-06,
-      "loss": 0.2322,
-      "step": 7000
-    },
-    {
-      "epoch": 11.551610239471511,
-      "eval_loss": 1.0933067798614502,
-      "eval_runtime": 721.5402,
-      "eval_samples_per_second": 14.664,
-      "eval_steps_per_second": 1.222,
-      "eval_wer": Infinity,
-      "step": 7000
-    },
-    {
-      "epoch": 11.59289843104872,
-      "grad_norm": 6.9122748374938965,
-      "learning_rate": 6.267368421052632e-06,
-      "loss": 0.2701,
-      "step": 7025
-    },
-    {
-      "epoch": 11.63418662262593,
-      "grad_norm": 4.586194038391113,
-      "learning_rate": 6.214736842105264e-06,
-      "loss": 0.2496,
-      "step": 7050
-    },
-    {
-      "epoch": 11.675474814203138,
-      "grad_norm": 5.90902853012085,
-      "learning_rate": 6.1621052631578945e-06,
-      "loss": 0.2384,
-      "step": 7075
-    },
-    {
-      "epoch": 11.716763005780347,
-      "grad_norm": 5.328728199005127,
-      "learning_rate": 6.109473684210527e-06,
-      "loss": 0.253,
-      "step": 7100
-    },
-    {
-      "epoch": 11.758051197357556,
-      "grad_norm": 6.127770900726318,
-      "learning_rate": 6.056842105263159e-06,
-      "loss": 0.2359,
-      "step": 7125
-    },
-    {
-      "epoch": 11.799339388934765,
-      "grad_norm": 5.675428867340088,
-      "learning_rate": 6.00421052631579e-06,
-      "loss": 0.2383,
-      "step": 7150
-    },
-    {
-      "epoch": 11.840627580511974,
-      "grad_norm": 4.921765327453613,
-      "learning_rate": 5.951578947368422e-06,
-      "loss": 0.2648,
-      "step": 7175
-    },
-    {
-      "epoch": 11.881915772089183,
-      "grad_norm": 4.198760986328125,
-      "learning_rate": 5.898947368421053e-06,
-      "loss": 0.2351,
-      "step": 7200
-    },
-    {
-      "epoch": 11.923203963666392,
-      "grad_norm": 4.786856174468994,
-      "learning_rate": 5.846315789473684e-06,
-      "loss": 0.2468,
-      "step": 7225
-    },
-    {
-      "epoch": 11.964492155243601,
-      "grad_norm": 4.955503463745117,
-      "learning_rate": 5.793684210526316e-06,
-      "loss": 0.2471,
-      "step": 7250
-    },
-    {
-      "epoch": 12.004954582989265,
-      "grad_norm": 4.753754138946533,
-      "learning_rate": 5.7410526315789485e-06,
-      "loss": 0.2625,
-      "step": 7275
-    },
-    {
-      "epoch": 12.046242774566474,
-      "grad_norm": 3.7155563831329346,
-      "learning_rate": 5.68842105263158e-06,
-      "loss": 0.2123,
-      "step": 7300
-    },
-    {
-      "epoch": 12.087530966143683,
-      "grad_norm": 4.637264251708984,
-      "learning_rate": 5.635789473684211e-06,
-      "loss": 0.209,
-      "step": 7325
-    },
-    {
-      "epoch": 12.128819157720892,
-      "grad_norm": 5.065235614776611,
-      "learning_rate": 5.5831578947368425e-06,
-      "loss": 0.2016,
-      "step": 7350
-    },
-    {
-      "epoch": 12.170107349298101,
-      "grad_norm": 4.761934280395508,
-      "learning_rate": 5.530526315789474e-06,
-      "loss": 0.2081,
-      "step": 7375
-    },
-    {
-      "epoch": 12.21139554087531,
-      "grad_norm": 5.17978048324585,
-      "learning_rate": 5.477894736842105e-06,
-      "loss": 0.2288,
-      "step": 7400
-    },
-    {
-      "epoch": 12.25268373245252,
-      "grad_norm": 3.554265260696411,
-      "learning_rate": 5.425263157894737e-06,
-      "loss": 0.2143,
-      "step": 7425
-    },
-    {
-      "epoch": 12.293971924029728,
-      "grad_norm": 5.236491680145264,
-      "learning_rate": 5.372631578947369e-06,
-      "loss": 0.2025,
-      "step": 7450
-    },
-    {
-      "epoch": 12.335260115606937,
-      "grad_norm": 5.198639392852783,
-      "learning_rate": 5.320000000000001e-06,
-      "loss": 0.2199,
-      "step": 7475
-    },
-    {
-      "epoch": 12.376548307184144,
-      "grad_norm": 5.509646892547607,
-      "learning_rate": 5.267368421052632e-06,
-      "loss": 0.1932,
-      "step": 7500
-    },
-    {
-      "epoch": 12.417836498761353,
-      "grad_norm": 4.569394588470459,
-      "learning_rate": 5.214736842105263e-06,
-      "loss": 0.2018,
-      "step": 7525
-    },
-    {
-      "epoch": 12.459124690338562,
-      "grad_norm": 4.686092376708984,
-      "learning_rate": 5.162105263157895e-06,
-      "loss": 0.2148,
-      "step": 7550
-    },
-    {
-      "epoch": 12.500412881915771,
-      "grad_norm": 5.779174327850342,
-      "learning_rate": 5.1094736842105264e-06,
-      "loss": 0.2362,
-      "step": 7575
-    },
-    {
-      "epoch": 12.54170107349298,
-      "grad_norm": 7.2701873779296875,
-      "learning_rate": 5.056842105263158e-06,
-      "loss": 0.2421,
-      "step": 7600
-    },
-    {
-      "epoch": 12.58298926507019,
-      "grad_norm": 4.832854270935059,
-      "learning_rate": 5.0042105263157906e-06,
-      "loss": 0.219,
-      "step": 7625
-    },
-    {
-      "epoch": 12.624277456647398,
-      "grad_norm": 5.803878307342529,
-      "learning_rate": 4.951578947368421e-06,
-      "loss": 0.2172,
-      "step": 7650
-    },
-    {
-      "epoch": 12.665565648224607,
-      "grad_norm": 4.021119594573975,
-      "learning_rate": 4.898947368421053e-06,
-      "loss": 0.227,
-      "step": 7675
-    },
-    {
-      "epoch": 12.706853839801816,
-      "grad_norm": 5.169255256652832,
-      "learning_rate": 4.846315789473685e-06,
-      "loss": 0.21,
-      "step": 7700
-    },
-    {
-      "epoch": 12.748142031379025,
-      "grad_norm": 4.798264503479004,
-      "learning_rate": 4.793684210526316e-06,
-      "loss": 0.1984,
-      "step": 7725
-    },
-    {
-      "epoch": 12.789430222956234,
-      "grad_norm": 5.631106853485107,
-      "learning_rate": 4.741052631578948e-06,
-      "loss": 0.225,
-      "step": 7750
-    },
-    {
-      "epoch": 12.830718414533443,
-      "grad_norm": 4.223946571350098,
-      "learning_rate": 4.6884210526315795e-06,
-      "loss": 0.2163,
-      "step": 7775
-    },
-    {
-      "epoch": 12.872006606110652,
-      "grad_norm": 4.958042621612549,
-      "learning_rate": 4.63578947368421e-06,
-      "loss": 0.2131,
-      "step": 7800
-    },
-    {
-      "epoch": 12.913294797687861,
-      "grad_norm": 6.063140392303467,
-      "learning_rate": 4.583157894736843e-06,
-      "loss": 0.2443,
-      "step": 7825
-    },
-    {
-      "epoch": 12.95458298926507,
-      "grad_norm": 4.037540435791016,
-      "learning_rate": 4.5305263157894744e-06,
-      "loss": 0.2033,
-      "step": 7850
-    },
-    {
-      "epoch": 12.995871180842279,
-      "grad_norm": 4.495209217071533,
-      "learning_rate": 4.477894736842105e-06,
-      "loss": 0.2307,
-      "step": 7875
-    },
-    {
-      "epoch": 13.036333608587944,
-      "grad_norm": 4.494256496429443,
-      "learning_rate": 4.425263157894737e-06,
-      "loss": 0.1894,
-      "step": 7900
-    },
-    {
-      "epoch": 13.077621800165153,
-      "grad_norm": 3.877058744430542,
-      "learning_rate": 4.3726315789473685e-06,
-      "loss": 0.195,
-      "step": 7925
-    },
-    {
-      "epoch": 13.118909991742361,
-      "grad_norm": 4.717246055603027,
-      "learning_rate": 4.32e-06,
-      "loss": 0.1745,
-      "step": 7950
-    },
-    {
-      "epoch": 13.16019818331957,
-      "grad_norm": 4.590612411499023,
-      "learning_rate": 4.267368421052632e-06,
-      "loss": 0.195,
-      "step": 7975
-    },
-    {
-      "epoch": 13.20148637489678,
-      "grad_norm": 6.558470249176025,
-      "learning_rate": 4.214736842105263e-06,
-      "loss": 0.1913,
-      "step": 8000
-    },
-    {
-      "epoch": 13.20148637489678,
-      "eval_loss": 1.1226582527160645,
-      "eval_runtime": 703.5225,
-      "eval_samples_per_second": 15.04,
-      "eval_steps_per_second": 1.254,
-      "eval_wer": Infinity,
-      "step": 8000
-    },
-    {
-      "epoch": 13.242774566473988,
-      "grad_norm": 4.316694736480713,
-      "learning_rate": 4.162105263157895e-06,
-      "loss": 0.2075,
-      "step": 8025
-    },
-    {
-      "epoch": 13.284062758051197,
-      "grad_norm": 4.108456611633301,
-      "learning_rate": 4.109473684210527e-06,
-      "loss": 0.1654,
-      "step": 8050
-    },
-    {
-      "epoch": 13.325350949628406,
-      "grad_norm": 3.923902750015259,
-      "learning_rate": 4.056842105263158e-06,
-      "loss": 0.19,
-      "step": 8075
-    },
-    {
-      "epoch": 13.366639141205615,
-      "grad_norm": 3.792008638381958,
-      "learning_rate": 4.00421052631579e-06,
-      "loss": 0.1924,
-      "step": 8100
-    },
-    {
-      "epoch": 13.407927332782824,
-      "grad_norm": 4.64762020111084,
-      "learning_rate": 3.951578947368422e-06,
-      "loss": 0.1988,
-      "step": 8125
-    },
-    {
-      "epoch": 13.449215524360033,
-      "grad_norm": 4.5705246925354,
-      "learning_rate": 3.898947368421052e-06,
-      "loss": 0.1857,
-      "step": 8150
-    },
-    {
-      "epoch": 13.490503715937242,
-      "grad_norm": 4.8831610679626465,
-      "learning_rate": 3.846315789473685e-06,
-      "loss": 0.1813,
-      "step": 8175
-    },
-    {
-      "epoch": 13.531791907514451,
-      "grad_norm": 3.7012646198272705,
-      "learning_rate": 3.793684210526316e-06,
-      "loss": 0.212,
-      "step": 8200
-    },
-    {
-      "epoch": 13.57308009909166,
-      "grad_norm": 4.698606967926025,
-      "learning_rate": 3.7410526315789473e-06,
-      "loss": 0.1999,
-      "step": 8225
-    },
-    {
-      "epoch": 13.61436829066887,
-      "grad_norm": 5.355463027954102,
-      "learning_rate": 3.6884210526315794e-06,
-      "loss": 0.188,
-      "step": 8250
-    },
-    {
-      "epoch": 13.655656482246078,
-      "grad_norm": 4.198258876800537,
-      "learning_rate": 3.635789473684211e-06,
-      "loss": 0.187,
-      "step": 8275
-    },
-    {
-      "epoch": 13.696944673823287,
-      "grad_norm": 4.937903881072998,
-      "learning_rate": 3.5831578947368422e-06,
-      "loss": 0.1942,
-      "step": 8300
-    },
-    {
-      "epoch": 13.738232865400496,
-      "grad_norm": 5.5391645431518555,
-      "learning_rate": 3.5305263157894743e-06,
-      "loss": 0.1897,
-      "step": 8325
-    },
-    {
-      "epoch": 13.779521056977705,
-      "grad_norm": 3.7583065032958984,
-      "learning_rate": 3.4778947368421055e-06,
-      "loss": 0.1932,
-      "step": 8350
-    },
-    {
-      "epoch": 13.820809248554914,
-      "grad_norm": 5.243954181671143,
-      "learning_rate": 3.425263157894737e-06,
-      "loss": 0.198,
-      "step": 8375
-    },
-    {
-      "epoch": 13.862097440132121,
-      "grad_norm": 5.0445556640625,
-      "learning_rate": 3.3726315789473683e-06,
-      "loss": 0.1773,
-      "step": 8400
-    },
-    {
-      "epoch": 13.90338563170933,
-      "grad_norm": 4.348180770874023,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 0.1834,
-      "step": 8425
-    },
-    {
-      "epoch": 13.94467382328654,
-      "grad_norm": 5.649644374847412,
-      "learning_rate": 3.267368421052632e-06,
-      "loss": 0.1772,
-      "step": 8450
-    },
-    {
-      "epoch": 13.985962014863748,
-      "grad_norm": 5.276495933532715,
-      "learning_rate": 3.2147368421052633e-06,
-      "loss": 0.211,
-      "step": 8475
-    },
-    {
-      "epoch": 14.026424442609414,
-      "grad_norm": 4.379940032958984,
-      "learning_rate": 3.1621052631578953e-06,
-      "loss": 0.1697,
-      "step": 8500
-    },
-    {
-      "epoch": 14.067712634186623,
-      "grad_norm": 4.398263454437256,
-      "learning_rate": 3.1094736842105265e-06,
-      "loss": 0.1656,
-      "step": 8525
-    },
-    {
-      "epoch": 14.109000825763832,
-      "grad_norm": 5.636984348297119,
-      "learning_rate": 3.056842105263158e-06,
-      "loss": 0.1613,
-      "step": 8550
-    },
-    {
-      "epoch": 14.15028901734104,
-      "grad_norm": 5.673081398010254,
-      "learning_rate": 3.0042105263157894e-06,
-      "loss": 0.1814,
-      "step": 8575
-    },
-    {
-      "epoch": 14.191577208918249,
-      "grad_norm": 4.340714454650879,
-      "learning_rate": 2.9515789473684214e-06,
-      "loss": 0.1482,
-      "step": 8600
-    },
-    {
-      "epoch": 14.232865400495458,
-      "grad_norm": 4.788120269775391,
-      "learning_rate": 2.8989473684210526e-06,
-      "loss": 0.1688,
-      "step": 8625
-    },
-    {
-      "epoch": 14.274153592072667,
-      "grad_norm": 4.258752822875977,
-      "learning_rate": 2.8463157894736843e-06,
-      "loss": 0.1623,
-      "step": 8650
-    },
-    {
-      "epoch": 14.315441783649876,
-      "grad_norm": 5.6167988777160645,
-      "learning_rate": 2.7936842105263163e-06,
-      "loss": 0.1736,
-      "step": 8675
-    },
-    {
-      "epoch": 14.356729975227084,
-      "grad_norm": 4.973904609680176,
-      "learning_rate": 2.7410526315789476e-06,
-      "loss": 0.163,
-      "step": 8700
-    },
-    {
-      "epoch": 14.398018166804293,
-      "grad_norm": 5.393108367919922,
-      "learning_rate": 2.688421052631579e-06,
-      "loss": 0.1721,
-      "step": 8725
-    },
-    {
-      "epoch": 14.439306358381502,
-      "grad_norm": 3.455310583114624,
-      "learning_rate": 2.635789473684211e-06,
-      "loss": 0.1762,
-      "step": 8750
-    },
-    {
-      "epoch": 14.480594549958711,
-      "grad_norm": 4.485291957855225,
-      "learning_rate": 2.5831578947368425e-06,
-      "loss": 0.167,
-      "step": 8775
-    },
-    {
-      "epoch": 14.52188274153592,
-      "grad_norm": 5.023384094238281,
-      "learning_rate": 2.5305263157894737e-06,
-      "loss": 0.1797,
-      "step": 8800
-    },
-    {
-      "epoch": 14.56317093311313,
-      "grad_norm": 3.6772608757019043,
-      "learning_rate": 2.4778947368421053e-06,
-      "loss": 0.1688,
-      "step": 8825
-    },
-    {
-      "epoch": 14.604459124690338,
-      "grad_norm": 4.417513847351074,
-      "learning_rate": 2.425263157894737e-06,
-      "loss": 0.176,
-      "step": 8850
-    },
-    {
-      "epoch": 14.645747316267547,
-      "grad_norm": 5.2892231941223145,
-      "learning_rate": 2.3726315789473686e-06,
-      "loss": 0.1802,
-      "step": 8875
-    },
-    {
-      "epoch": 14.687035507844756,
-      "grad_norm": 5.4643025398254395,
-      "learning_rate": 2.3200000000000002e-06,
-      "loss": 0.1914,
-      "step": 8900
-    },
-    {
-      "epoch": 14.728323699421965,
-      "grad_norm": 3.4271011352539062,
-      "learning_rate": 2.267368421052632e-06,
-      "loss": 0.1668,
-      "step": 8925
-    },
-    {
-      "epoch": 14.769611890999174,
-      "grad_norm": 4.431731224060059,
-      "learning_rate": 2.2147368421052635e-06,
-      "loss": 0.1725,
-      "step": 8950
-    },
-    {
-      "epoch": 14.810900082576383,
-      "grad_norm": 3.5097200870513916,
-      "learning_rate": 2.1621052631578947e-06,
-      "loss": 0.169,
-      "step": 8975
-    },
-    {
-      "epoch": 14.852188274153592,
-      "grad_norm": 4.471765995025635,
-      "learning_rate": 2.1094736842105264e-06,
-      "loss": 0.1424,
-      "step": 9000
-    },
-    {
-      "epoch": 14.852188274153592,
-      "eval_loss": 1.1418445110321045,
-      "eval_runtime": 697.6196,
-      "eval_samples_per_second": 15.167,
-      "eval_steps_per_second": 1.264,
-      "eval_wer": Infinity,
-      "step": 9000
-    },
-    {
-      "epoch": 14.893476465730801,
-      "grad_norm": 4.103869915008545,
-      "learning_rate": 2.056842105263158e-06,
-      "loss": 0.1855,
-      "step": 9025
-    },
-    {
-      "epoch": 14.93476465730801,
-      "grad_norm": 4.589334487915039,
-      "learning_rate": 2.0042105263157896e-06,
-      "loss": 0.1723,
-      "step": 9050
-    },
-    {
-      "epoch": 14.97605284888522,
-      "grad_norm": 5.06859016418457,
-      "learning_rate": 1.9515789473684213e-06,
-      "loss": 0.1568,
-      "step": 9075
-    },
-    {
-      "epoch": 15.016515276630884,
-      "grad_norm": 3.8599202632904053,
-      "learning_rate": 1.8989473684210527e-06,
-      "loss": 0.1802,
-      "step": 9100
-    },
-    {
-      "epoch": 15.057803468208093,
-      "grad_norm": 3.51918363571167,
-      "learning_rate": 1.8463157894736843e-06,
-      "loss": 0.163,
-      "step": 9125
-    },
-    {
-      "epoch": 15.099091659785302,
-      "grad_norm": 3.9094550609588623,
-      "learning_rate": 1.7936842105263158e-06,
-      "loss": 0.1508,
-      "step": 9150
-    },
-    {
-      "epoch": 15.14037985136251,
-      "grad_norm": 4.491118907928467,
-      "learning_rate": 1.7410526315789474e-06,
-      "loss": 0.1683,
-      "step": 9175
-    },
-    {
-      "epoch": 15.18166804293972,
-      "grad_norm": 5.513247966766357,
-      "learning_rate": 1.6884210526315792e-06,
-      "loss": 0.1413,
-      "step": 9200
-    },
-    {
-      "epoch": 15.222956234516928,
-      "grad_norm": 4.3536577224731445,
-      "learning_rate": 1.6357894736842107e-06,
-      "loss": 0.1749,
-      "step": 9225
-    },
-    {
-      "epoch": 15.264244426094137,
-      "grad_norm": 3.8934166431427,
-      "learning_rate": 1.5831578947368423e-06,
-      "loss": 0.1727,
-      "step": 9250
-    },
-    {
-      "epoch": 15.305532617671346,
-      "grad_norm": 4.610677242279053,
-      "learning_rate": 1.5305263157894737e-06,
-      "loss": 0.1518,
-      "step": 9275
-    },
-    {
-      "epoch": 15.346820809248555,
-      "grad_norm": 3.230677366256714,
-      "learning_rate": 1.4778947368421054e-06,
-      "loss": 0.1525,
-      "step": 9300
-    },
-    {
-      "epoch": 15.388109000825764,
-      "grad_norm": 3.854262590408325,
-      "learning_rate": 1.425263157894737e-06,
-      "loss": 0.1501,
-      "step": 9325
-    },
-    {
-      "epoch": 15.429397192402973,
-      "grad_norm": 3.96171498298645,
-      "learning_rate": 1.3726315789473684e-06,
-      "loss": 0.1655,
-      "step": 9350
-    },
-    {
-      "epoch": 15.470685383980182,
-      "grad_norm": 3.8165667057037354,
-      "learning_rate": 1.32e-06,
-      "loss": 0.1469,
-      "step": 9375
-    },
-    {
-      "epoch": 15.511973575557391,
-      "grad_norm": 3.6944658756256104,
-      "learning_rate": 1.2673684210526315e-06,
-      "loss": 0.15,
-      "step": 9400
-    },
-    {
-      "epoch": 15.5532617671346,
-      "grad_norm": 3.9097495079040527,
-      "learning_rate": 1.2147368421052633e-06,
-      "loss": 0.1492,
-      "step": 9425
-    },
-    {
-      "epoch": 15.59454995871181,
-      "grad_norm": 5.010981559753418,
-      "learning_rate": 1.1621052631578948e-06,
-      "loss": 0.1551,
-      "step": 9450
-    },
-    {
-      "epoch": 15.635838150289018,
-      "grad_norm": 4.151124000549316,
-      "learning_rate": 1.1094736842105264e-06,
-      "loss": 0.1569,
-      "step": 9475
-    },
-    {
-      "epoch": 15.677126341866225,
-      "grad_norm": 4.455909729003906,
-      "learning_rate": 1.0568421052631578e-06,
-      "loss": 0.1551,
-      "step": 9500
-    },
-    {
-      "epoch": 15.718414533443434,
-      "grad_norm": 4.5624895095825195,
-      "learning_rate": 1.0042105263157897e-06,
-      "loss": 0.1526,
-      "step": 9525
-    },
-    {
-      "epoch": 15.759702725020643,
-      "grad_norm": 3.476386070251465,
-      "learning_rate": 9.515789473684212e-07,
-      "loss": 0.151,
-      "step": 9550
-    },
-    {
-      "epoch": 15.800990916597852,
-      "grad_norm": 3.7594966888427734,
-      "learning_rate": 8.989473684210527e-07,
-      "loss": 0.1614,
-      "step": 9575
-    },
-    {
-      "epoch": 15.842279108175061,
-      "grad_norm": 3.677940607070923,
-      "learning_rate": 8.463157894736843e-07,
-      "loss": 0.1607,
-      "step": 9600
-    },
-    {
-      "epoch": 15.88356729975227,
-      "grad_norm": 2.814276695251465,
-      "learning_rate": 7.936842105263158e-07,
-      "loss": 0.1584,
-      "step": 9625
-    },
-    {
-      "epoch": 15.92485549132948,
-      "grad_norm": 4.551584243774414,
-      "learning_rate": 7.410526315789475e-07,
-      "loss": 0.1746,
-      "step": 9650
-    },
-    {
-      "epoch": 15.966143682906688,
-      "grad_norm": 3.54571533203125,
-      "learning_rate": 6.884210526315791e-07,
-      "loss": 0.1283,
-      "step": 9675
-    },
-    {
-      "epoch": 16.006606110652353,
-      "grad_norm": 3.034546136856079,
-      "learning_rate": 6.357894736842106e-07,
-      "loss": 0.1445,
-      "step": 9700
-    },
-    {
-      "epoch": 16.04789430222956,
-      "grad_norm": 4.239781379699707,
-      "learning_rate": 5.831578947368421e-07,
-      "loss": 0.1526,
-      "step": 9725
-    },
-    {
-      "epoch": 16.08918249380677,
-      "grad_norm": 3.5759620666503906,
-      "learning_rate": 5.305263157894737e-07,
-      "loss": 0.1435,
-      "step": 9750
-    },
-    {
-      "epoch": 16.13047068538398,
-      "grad_norm": 3.807652473449707,
-      "learning_rate": 4.778947368421053e-07,
-      "loss": 0.1616,
-      "step": 9775
-    },
-    {
-      "epoch": 16.17175887696119,
-      "grad_norm": 2.9478118419647217,
-      "learning_rate": 4.2526315789473684e-07,
-      "loss": 0.1344,
-      "step": 9800
-    },
-    {
-      "epoch": 16.213047068538398,
-      "grad_norm": 3.351928472518921,
-      "learning_rate": 3.726315789473685e-07,
-      "loss": 0.1616,
-      "step": 9825
-    },
-    {
-      "epoch": 16.254335260115607,
-      "grad_norm": 4.71775484085083,
-      "learning_rate": 3.2e-07,
-      "loss": 0.138,
-      "step": 9850
-    },
-    {
-      "epoch": 16.295623451692816,
-      "grad_norm": 4.4986138343811035,
-      "learning_rate": 2.6736842105263164e-07,
-      "loss": 0.1592,
-      "step": 9875
-    },
-    {
-      "epoch": 16.336911643270025,
-      "grad_norm": 4.285621643066406,
-      "learning_rate": 2.1473684210526317e-07,
-      "loss": 0.1423,
-      "step": 9900
-    },
-    {
-      "epoch": 16.378199834847234,
-      "grad_norm": 4.092608451843262,
-      "learning_rate": 1.6210526315789476e-07,
-      "loss": 0.1619,
-      "step": 9925
-    },
-    {
-      "epoch": 16.419488026424442,
-      "grad_norm": 4.384539604187012,
-      "learning_rate": 1.0947368421052632e-07,
-      "loss": 0.1606,
-      "step": 9950
-    },
-    {
-      "epoch": 16.46077621800165,
-      "grad_norm": 5.559605121612549,
-      "learning_rate": 5.68421052631579e-08,
-      "loss": 0.164,
-      "step": 9975
-    },
-    {
-      "epoch": 16.50206440957886,
-      "grad_norm": 4.197207450866699,
-      "learning_rate": 4.210526315789474e-09,
-      "loss": 0.1615,
-      "step": 10000
-    },
-    {
-      "epoch": 16.50206440957886,
-      "eval_loss": 1.1515085697174072,
-      "eval_runtime": 702.9423,
-      "eval_samples_per_second": 15.052,
-      "eval_steps_per_second": 1.255,
-      "eval_wer": Infinity,
-      "step": 10000
     }
   ],
   "logging_steps": 25,
@@ -2911,12 +311,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 1.949150849531904e+19,
+  "total_flos": 1.94993134239744e+18,
   "train_batch_size": 12,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86669d9379dd46ebaf27001a30f388cf74da12189f00c1f8f129de756214b0bb
+oid sha256:f8fe144e7bbecfefafbc5f7ba6e25af714f15b9643283e39c2a95c6ef7dcd775
 size 5560