Upload 13 files

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1aacd357196d2ccb7b69b7eabae19fa89e889c9564d29e8e8552c19f401158b0
+oid sha256:2e622da60d23e94e07cdb99ff234aaefa76f7c6fb21c57c6ab61ba5d19f8b9c7
 size 2490594117

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea7b88a67ffa34b8ff340a18a3f92eda4b2cd4806c641c0324b8c8995cca408d
+oid sha256:a4512cd385771733271cfd87d2c72759096c7a1a777295cc9b7fef89f555dfd2
 size 1262168365

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:811f2ca8b1e9e9b61905e518ff2bbfb86d07420b0db70afcaa4bb789c5fc7609
+oid sha256:015a8d4a0507b7ea92a5d2b6e83f85108572344d34478a240dd0c974135c5e09
 size 14575

scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b75f99a5a4a4b6ae83f76610c6becc08f675166496a0e823528855b7dc030640
+oid sha256:db575b4427150600ee37f6c6a1a316680eb05224eb75e119839f9b6fbad60117
 size 557

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:11fb4e3697e269e34809008650d5c221e8a285a7472a4da3e755dee9bf0e7da6
+oid sha256:7ff0fa787e174ecf06cbb98b01670eaefa203dc8ba7e6a8d2ed74e188444dfb3
 size 627

trainer_state.json

@@ -1,1771 +1,616 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 95.90163934426229,
-  "global_step": 11700,
+  "epoch": 31.99304865938431,
+  "global_step": 4000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.82,
+      "epoch": 0.79,
       "learning_rate": 0.0001,
-      "loss": 0.0216,
+      "loss": 0.0392,
       "step": 100
     },
     {
-      "epoch": 0.82,
-      "eval_loss": 0.31533992290496826,
-      "eval_runtime": 207.6066,
-      "eval_samples_per_second": 16.782,
-      "eval_steps_per_second": 2.1,
-      "eval_wer": 0.21548575602629658,
+      "epoch": 0.79,
+      "eval_loss": 0.23765751719474792,
+      "eval_runtime": 220.5891,
+      "eval_samples_per_second": 16.297,
+      "eval_steps_per_second": 2.04,
+      "eval_wer": 0.21887876816308827,
       "step": 100
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 9.991796554552912e-05,
-      "loss": 0.0195,
+      "epoch": 1.6,
+      "learning_rate": 9.991993594875901e-05,
+      "loss": 0.0336,
       "step": 200
     },
     {
-      "epoch": 1.64,
-      "eval_loss": 0.33585426211357117,
-      "eval_runtime": 141.4442,
-      "eval_samples_per_second": 24.632,
-      "eval_steps_per_second": 3.082,
-      "eval_wer": 0.21469910659099847,
+      "epoch": 1.6,
+      "eval_loss": 0.26165521144866943,
+      "eval_runtime": 157.9726,
+      "eval_samples_per_second": 22.757,
+      "eval_steps_per_second": 2.849,
+      "eval_wer": 0.21649316851008457,
       "step": 200
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 9.983593109105825e-05,
-      "loss": 0.0213,
+      "epoch": 2.4,
+      "learning_rate": 9.983987189751803e-05,
+      "loss": 0.0293,
       "step": 300
     },
     {
-      "epoch": 2.46,
-      "eval_loss": 0.32860827445983887,
-      "eval_runtime": 143.9808,
-      "eval_samples_per_second": 24.198,
-      "eval_steps_per_second": 3.028,
-      "eval_wer": 0.21891329999438108,
+      "epoch": 2.4,
+      "eval_loss": 0.28323182463645935,
+      "eval_runtime": 169.3448,
+      "eval_samples_per_second": 21.229,
+      "eval_steps_per_second": 2.657,
+      "eval_wer": 0.2197462589459987,
       "step": 300
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 9.975389663658737e-05,
-      "loss": 0.0207,
+      "epoch": 3.2,
+      "learning_rate": 9.975980784627703e-05,
+      "loss": 0.0283,
       "step": 400
     },
     {
-      "epoch": 3.28,
-      "eval_loss": 0.34261101484298706,
-      "eval_runtime": 151.928,
-      "eval_samples_per_second": 22.932,
-      "eval_steps_per_second": 2.87,
-      "eval_wer": 0.21542956678091812,
+      "epoch": 3.2,
+      "eval_loss": 0.29311421513557434,
+      "eval_runtime": 175.6178,
+      "eval_samples_per_second": 20.471,
+      "eval_steps_per_second": 2.562,
+      "eval_wer": 0.21508349598785512,
       "step": 400
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 9.967186218211649e-05,
-      "loss": 0.0218,
+      "epoch": 3.99,
+      "learning_rate": 9.967974379503603e-05,
+      "loss": 0.0274,
       "step": 500
     },
     {
-      "epoch": 4.1,
-      "eval_loss": 0.3265298306941986,
-      "eval_runtime": 157.208,
-      "eval_samples_per_second": 22.162,
-      "eval_steps_per_second": 2.773,
-      "eval_wer": 0.2182952182952183,
+      "epoch": 3.99,
+      "eval_loss": 0.30566948652267456,
+      "eval_runtime": 182.1945,
+      "eval_samples_per_second": 19.732,
+      "eval_steps_per_second": 2.47,
+      "eval_wer": 0.21828236824983735,
       "step": 500
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 9.958982772764562e-05,
-      "loss": 0.0197,
+      "epoch": 4.79,
+      "learning_rate": 9.959967974379504e-05,
+      "loss": 0.0264,
       "step": 600
     },
     {
-      "epoch": 4.92,
-      "eval_loss": 0.3343443274497986,
-      "eval_runtime": 155.21,
-      "eval_samples_per_second": 22.447,
-      "eval_steps_per_second": 2.809,
-      "eval_wer": 0.22161038377254594,
+      "epoch": 4.79,
+      "eval_loss": 0.30197781324386597,
+      "eval_runtime": 174.7969,
+      "eval_samples_per_second": 20.567,
+      "eval_steps_per_second": 2.574,
+      "eval_wer": 0.2171437865972674,
       "step": 600
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 9.950779327317475e-05,
-      "loss": 0.0199,
+      "epoch": 5.6,
+      "learning_rate": 9.951961569255405e-05,
+      "loss": 0.0259,
       "step": 700
     },
     {
-      "epoch": 5.74,
-      "eval_loss": 0.34997832775115967,
-      "eval_runtime": 162.2341,
-      "eval_samples_per_second": 21.475,
-      "eval_steps_per_second": 2.687,
-      "eval_wer": 0.21885711074900263,
+      "epoch": 5.6,
+      "eval_loss": 0.3002428412437439,
+      "eval_runtime": 191.2838,
+      "eval_samples_per_second": 18.794,
+      "eval_steps_per_second": 2.353,
+      "eval_wer": 0.21725222294513122,
       "step": 700
     },
     {
-      "epoch": 6.56,
-      "learning_rate": 9.942575881870387e-05,
-      "loss": 0.0208,
+      "epoch": 6.4,
+      "learning_rate": 9.943955164131305e-05,
+      "loss": 0.0254,
       "step": 800
     },
     {
-      "epoch": 6.56,
-      "eval_loss": 0.3275943994522095,
-      "eval_runtime": 159.135,
-      "eval_samples_per_second": 21.893,
-      "eval_steps_per_second": 2.74,
-      "eval_wer": 0.22048659886497723,
+      "epoch": 6.4,
+      "eval_loss": 0.3097267746925354,
+      "eval_runtime": 183.9687,
+      "eval_samples_per_second": 19.541,
+      "eval_steps_per_second": 2.446,
+      "eval_wer": 0.21752331381479073,
       "step": 800
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 9.934372436423298e-05,
-      "loss": 0.0206,
+      "epoch": 7.2,
+      "learning_rate": 9.936028823058447e-05,
+      "loss": 0.0252,
       "step": 900
     },
     {
-      "epoch": 7.38,
-      "eval_loss": 0.3496840000152588,
-      "eval_runtime": 160.165,
-      "eval_samples_per_second": 21.753,
-      "eval_steps_per_second": 2.722,
-      "eval_wer": 0.21329437545653762,
+      "epoch": 7.2,
+      "eval_loss": 0.2970833480358124,
+      "eval_runtime": 182.1563,
+      "eval_samples_per_second": 19.736,
+      "eval_steps_per_second": 2.47,
+      "eval_wer": 0.2170353502494036,
       "step": 900
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 9.92616899097621e-05,
-      "loss": 0.0195,
+      "epoch": 7.99,
+      "learning_rate": 9.928022417934349e-05,
+      "loss": 0.0234,
       "step": 1000
     },
     {
-      "epoch": 8.2,
-      "eval_loss": 0.3464973568916321,
-      "eval_runtime": 159.4572,
-      "eval_samples_per_second": 21.849,
-      "eval_steps_per_second": 2.734,
-      "eval_wer": 0.22172276226330281,
+      "epoch": 7.99,
+      "eval_loss": 0.31429246068000793,
+      "eval_runtime": 193.7999,
+      "eval_samples_per_second": 18.55,
+      "eval_steps_per_second": 2.322,
+      "eval_wer": 0.21405335068314899,
       "step": 1000
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 9.917965545529123e-05,
-      "loss": 0.0215,
+      "epoch": 8.79,
+      "learning_rate": 9.920016012810248e-05,
+      "loss": 0.0228,
       "step": 1100
     },
     {
-      "epoch": 9.02,
-      "eval_loss": 0.34134212136268616,
-      "eval_runtime": 161.922,
-      "eval_samples_per_second": 21.517,
-      "eval_steps_per_second": 2.693,
-      "eval_wer": 0.22599314491206382,
+      "epoch": 8.79,
+      "eval_loss": 0.32985326647758484,
+      "eval_runtime": 194.3544,
+      "eval_samples_per_second": 18.497,
+      "eval_steps_per_second": 2.315,
+      "eval_wer": 0.2142702233788766,
       "step": 1100
     },
     {
-      "epoch": 9.84,
-      "learning_rate": 9.909762100082035e-05,
-      "loss": 0.0222,
+      "epoch": 9.6,
+      "learning_rate": 9.912009607686149e-05,
+      "loss": 0.0236,
       "step": 1200
     },
     {
-      "epoch": 9.84,
-      "eval_loss": 0.34097427129745483,
-      "eval_runtime": 163.193,
-      "eval_samples_per_second": 21.349,
-      "eval_steps_per_second": 2.672,
-      "eval_wer": 0.22852166095409338,
+      "epoch": 9.6,
+      "eval_loss": 0.3165593147277832,
+      "eval_runtime": 185.0001,
+      "eval_samples_per_second": 19.432,
+      "eval_steps_per_second": 2.432,
+      "eval_wer": 0.21833658642376924,
       "step": 1200
     },
     {
-      "epoch": 10.66,
-      "learning_rate": 9.901558654634947e-05,
-      "loss": 0.0214,
+      "epoch": 10.4,
+      "learning_rate": 9.90400320256205e-05,
+      "loss": 0.0241,
       "step": 1300
     },
     {
-      "epoch": 10.66,
-      "eval_loss": 0.3526441752910614,
-      "eval_runtime": 167.13,
-      "eval_samples_per_second": 20.846,
-      "eval_steps_per_second": 2.609,
-      "eval_wer": 0.2244760352868461,
+      "epoch": 10.4,
+      "eval_loss": 0.3285478949546814,
+      "eval_runtime": 181.9532,
+      "eval_samples_per_second": 19.758,
+      "eval_steps_per_second": 2.473,
+      "eval_wer": 0.2192582953806116,
       "step": 1300
     },
     {
-      "epoch": 11.48,
-      "learning_rate": 9.893355209187858e-05,
-      "loss": 0.0214,
+      "epoch": 11.2,
+      "learning_rate": 9.895996797437951e-05,
+      "loss": 0.0243,
       "step": 1400
     },
     {
-      "epoch": 11.48,
-      "eval_loss": 0.34132060408592224,
-      "eval_runtime": 193.5042,
-      "eval_samples_per_second": 18.005,
-      "eval_steps_per_second": 2.253,
-      "eval_wer": 0.2203742203742204,
+      "epoch": 11.2,
+      "eval_loss": 0.3187803626060486,
+      "eval_runtime": 200.4409,
+      "eval_samples_per_second": 17.935,
+      "eval_steps_per_second": 2.245,
+      "eval_wer": 0.22099327694643245,
       "step": 1400
     },
     {
-      "epoch": 12.3,
-      "learning_rate": 9.885151763740772e-05,
-      "loss": 0.0222,
+      "epoch": 11.99,
+      "learning_rate": 9.887990392313852e-05,
+      "loss": 0.026,
       "step": 1500
     },
     {
-      "epoch": 12.3,
-      "eval_loss": 0.34180501103401184,
-      "eval_runtime": 195.2175,
-      "eval_samples_per_second": 17.847,
-      "eval_steps_per_second": 2.233,
-      "eval_wer": 0.22211608698095184,
+      "epoch": 11.99,
+      "eval_loss": 0.32988375425338745,
+      "eval_runtime": 199.5224,
+      "eval_samples_per_second": 18.018,
+      "eval_steps_per_second": 2.255,
+      "eval_wer": 0.22375840381695944,
       "step": 1500
     },
     {
-      "epoch": 13.11,
-      "learning_rate": 9.876948318293683e-05,
-      "loss": 0.024,
+      "epoch": 12.79,
+      "learning_rate": 9.879983987189752e-05,
+      "loss": 0.0259,
       "step": 1600
     },
     {
-      "epoch": 13.11,
-      "eval_loss": 0.3383408486843109,
-      "eval_runtime": 197.1656,
-      "eval_samples_per_second": 17.67,
-      "eval_steps_per_second": 2.211,
-      "eval_wer": 0.22352081811541272,
+      "epoch": 12.79,
+      "eval_loss": 0.3099309206008911,
+      "eval_runtime": 198.0592,
+      "eval_samples_per_second": 18.151,
+      "eval_steps_per_second": 2.272,
+      "eval_wer": 0.22045109520711342,
       "step": 1600
     },
     {
-      "epoch": 13.93,
-      "learning_rate": 9.868744872846596e-05,
-      "loss": 0.0214,
+      "epoch": 13.6,
+      "learning_rate": 9.871977582065654e-05,
+      "loss": 0.0255,
       "step": 1700
     },
     {
-      "epoch": 13.93,
-      "eval_loss": 0.3334263563156128,
-      "eval_runtime": 193.741,
-      "eval_samples_per_second": 17.983,
-      "eval_steps_per_second": 2.25,
-      "eval_wer": 0.21632859470697308,
+      "epoch": 13.6,
+      "eval_loss": 0.30543622374534607,
+      "eval_runtime": 202.823,
+      "eval_samples_per_second": 17.725,
+      "eval_steps_per_second": 2.219,
+      "eval_wer": 0.22402949468661895,
       "step": 1700
     },
     {
-      "epoch": 14.75,
-      "learning_rate": 9.860541427399508e-05,
-      "loss": 0.02,
+      "epoch": 14.4,
+      "learning_rate": 9.863971176941553e-05,
+      "loss": 0.0253,
       "step": 1800
     },
     {
-      "epoch": 14.75,
-      "eval_loss": 0.336081862449646,
-      "eval_runtime": 196.0013,
-      "eval_samples_per_second": 17.775,
-      "eval_steps_per_second": 2.224,
-      "eval_wer": 0.2171714333876496,
+      "epoch": 14.4,
+      "eval_loss": 0.31324318051338196,
+      "eval_runtime": 199.664,
+      "eval_samples_per_second": 18.005,
+      "eval_steps_per_second": 2.254,
+      "eval_wer": 0.21779440468445022,
       "step": 1800
     },
     {
-      "epoch": 15.57,
-      "learning_rate": 9.852337981952421e-05,
-      "loss": 0.0196,
+      "epoch": 15.2,
+      "learning_rate": 9.855964771817454e-05,
+      "loss": 0.0244,
       "step": 1900
     },
     {
-      "epoch": 15.57,
-      "eval_loss": 0.33557742834091187,
-      "eval_runtime": 195.9355,
-      "eval_samples_per_second": 17.781,
-      "eval_steps_per_second": 2.225,
-      "eval_wer": 0.2231836826431421,
+      "epoch": 15.2,
+      "eval_loss": 0.33187857270240784,
+      "eval_runtime": 197.0859,
+      "eval_samples_per_second": 18.241,
+      "eval_steps_per_second": 2.283,
+      "eval_wer": 0.22121014964216004,
       "step": 1900
     },
     {
-      "epoch": 16.39,
-      "learning_rate": 9.844134536505333e-05,
-      "loss": 0.0192,
+      "epoch": 15.99,
+      "learning_rate": 9.847958366693355e-05,
+      "loss": 0.0231,
       "step": 2000
     },
     {
-      "epoch": 16.39,
-      "eval_loss": 0.34101295471191406,
-      "eval_runtime": 196.0796,
-      "eval_samples_per_second": 17.768,
-      "eval_steps_per_second": 2.224,
-      "eval_wer": 0.21891329999438108,
+      "epoch": 15.99,
+      "eval_loss": 0.33831512928009033,
+      "eval_runtime": 200.252,
+      "eval_samples_per_second": 17.952,
+      "eval_steps_per_second": 2.247,
+      "eval_wer": 0.21920407720667967,
       "step": 2000
     },
     {
-      "epoch": 17.21,
-      "learning_rate": 9.835931091058245e-05,
-      "loss": 0.0194,
+      "epoch": 16.79,
+      "learning_rate": 9.839951961569256e-05,
+      "loss": 0.0235,
       "step": 2100
     },
     {
-      "epoch": 17.21,
-      "eval_loss": 0.33082982897758484,
-      "eval_runtime": 197.1446,
-      "eval_samples_per_second": 17.672,
-      "eval_steps_per_second": 2.212,
-      "eval_wer": 0.22065516660111253,
+      "epoch": 16.79,
+      "eval_loss": 0.31389620900154114,
+      "eval_runtime": 198.2191,
+      "eval_samples_per_second": 18.136,
+      "eval_steps_per_second": 2.27,
+      "eval_wer": 0.21622207764042506,
       "step": 2100
     },
     {
-      "epoch": 18.03,
-      "learning_rate": 9.827727645611156e-05,
-      "loss": 0.0185,
+      "epoch": 17.6,
+      "learning_rate": 9.831945556445156e-05,
+      "loss": 0.0227,
       "step": 2200
     },
     {
-      "epoch": 18.03,
-      "eval_loss": 0.32813236117362976,
-      "eval_runtime": 196.5485,
-      "eval_samples_per_second": 17.726,
+      "epoch": 17.6,
+      "eval_loss": 0.32037118077278137,
+      "eval_runtime": 202.9198,
+      "eval_samples_per_second": 17.716,
       "eval_steps_per_second": 2.218,
-      "eval_wer": 0.21891329999438108,
+      "eval_wer": 0.217360659292995,
       "step": 2200
     },
     {
-      "epoch": 18.85,
-      "learning_rate": 9.81952420016407e-05,
-      "loss": 0.0193,
+      "epoch": 18.4,
+      "learning_rate": 9.823939151321058e-05,
+      "loss": 0.0228,
       "step": 2300
     },
     {
-      "epoch": 18.85,
-      "eval_loss": 0.33049654960632324,
-      "eval_runtime": 199.567,
-      "eval_samples_per_second": 17.458,
-      "eval_steps_per_second": 2.185,
-      "eval_wer": 0.22020565263808506,
+      "epoch": 18.4,
+      "eval_loss": 0.32169201970100403,
+      "eval_runtime": 193.9,
+      "eval_samples_per_second": 18.54,
+      "eval_steps_per_second": 2.321,
+      "eval_wer": 0.21757753198872262,
       "step": 2300
     },
     {
-      "epoch": 19.67,
-      "learning_rate": 9.811320754716981e-05,
-      "loss": 0.0187,
+      "epoch": 19.2,
+      "learning_rate": 9.815932746196959e-05,
+      "loss": 0.0217,
       "step": 2400
     },
     {
-      "epoch": 19.67,
-      "eval_loss": 0.32034164667129517,
-      "eval_runtime": 194.435,
-      "eval_samples_per_second": 17.919,
-      "eval_steps_per_second": 2.242,
-      "eval_wer": 0.2191380569758948,
+      "epoch": 19.2,
+      "eval_loss": 0.3112569749355316,
+      "eval_runtime": 198.2135,
+      "eval_samples_per_second": 18.137,
+      "eval_steps_per_second": 2.27,
+      "eval_wer": 0.2170895684233355,
       "step": 2400
     },
     {
-      "epoch": 20.49,
-      "learning_rate": 9.803117309269893e-05,
-      "loss": 0.0198,
+      "epoch": 19.99,
+      "learning_rate": 9.807926341072858e-05,
+      "loss": 0.0212,
       "step": 2500
     },
     {
-      "epoch": 20.49,
-      "eval_loss": 0.3243141174316406,
-      "eval_runtime": 195.3165,
-      "eval_samples_per_second": 17.838,
-      "eval_steps_per_second": 2.232,
-      "eval_wer": 0.21981232792043603,
+      "epoch": 19.99,
+      "eval_loss": 0.31596991419792175,
+      "eval_runtime": 199.2987,
+      "eval_samples_per_second": 18.038,
+      "eval_steps_per_second": 2.258,
+      "eval_wer": 0.21351116894382996,
       "step": 2500
     },
     {
-      "epoch": 21.31,
-      "learning_rate": 9.794995898277277e-05,
-      "loss": 0.02,
+      "epoch": 20.79,
+      "learning_rate": 9.79991993594876e-05,
+      "loss": 0.0216,
       "step": 2600
     },
     {
-      "epoch": 21.31,
-      "eval_loss": 0.34270626306533813,
-      "eval_runtime": 198.8904,
-      "eval_samples_per_second": 17.517,
-      "eval_steps_per_second": 2.192,
-      "eval_wer": 0.22082373433724786,
+      "epoch": 20.79,
+      "eval_loss": 0.3226161003112793,
+      "eval_runtime": 199.747,
+      "eval_samples_per_second": 17.998,
+      "eval_steps_per_second": 2.253,
+      "eval_wer": 0.21378225981348947,
       "step": 2600
     },
     {
-      "epoch": 22.13,
-      "learning_rate": 9.786792452830189e-05,
-      "loss": 0.0183,
+      "epoch": 21.6,
+      "learning_rate": 9.79191353082466e-05,
+      "loss": 0.0242,
       "step": 2700
     },
     {
-      "epoch": 22.13,
-      "eval_loss": 0.3502834141254425,
-      "eval_runtime": 185.3267,
-      "eval_samples_per_second": 18.799,
-      "eval_steps_per_second": 2.353,
-      "eval_wer": 0.21874473225824578,
+      "epoch": 21.6,
+      "eval_loss": 0.3281223177909851,
+      "eval_runtime": 197.1976,
+      "eval_samples_per_second": 18.23,
+      "eval_steps_per_second": 2.282,
+      "eval_wer": 0.2169269139015398,
       "step": 2700
     },
     {
-      "epoch": 22.95,
-      "learning_rate": 9.7785890073831e-05,
-      "loss": 0.0171,
+      "epoch": 22.4,
+      "learning_rate": 9.783907125700561e-05,
+      "loss": 0.0245,
       "step": 2800
     },
     {
-      "epoch": 22.95,
-      "eval_loss": 0.3397138714790344,
-      "eval_runtime": 200.1956,
-      "eval_samples_per_second": 17.403,
-      "eval_steps_per_second": 2.178,
-      "eval_wer": 0.22228465471708714,
+      "epoch": 22.4,
+      "eval_loss": 0.3078162968158722,
+      "eval_runtime": 198.6659,
+      "eval_samples_per_second": 18.096,
+      "eval_steps_per_second": 2.265,
+      "eval_wer": 0.2176317501626545,
       "step": 2800
     },
     {
-      "epoch": 23.77,
-      "learning_rate": 9.770385561936014e-05,
-      "loss": 0.018,
+      "epoch": 23.2,
+      "learning_rate": 9.775900720576461e-05,
+      "loss": 0.0229,
       "step": 2900
     },
     {
-      "epoch": 23.77,
-      "eval_loss": 0.3545791208744049,
-      "eval_runtime": 199.5635,
-      "eval_samples_per_second": 17.458,
-      "eval_steps_per_second": 2.185,
-      "eval_wer": 0.2228465471708715,
+      "epoch": 23.2,
+      "eval_loss": 0.31995320320129395,
+      "eval_runtime": 204.328,
+      "eval_samples_per_second": 17.594,
+      "eval_steps_per_second": 2.202,
+      "eval_wer": 0.22045109520711342,
       "step": 2900
     },
     {
-      "epoch": 24.59,
-      "learning_rate": 9.762182116488927e-05,
-      "loss": 0.0175,
+      "epoch": 23.99,
+      "learning_rate": 9.767894315452363e-05,
+      "loss": 0.0226,
       "step": 3000
     },
     {
-      "epoch": 24.59,
-      "eval_loss": 0.33466094732284546,
-      "eval_runtime": 198.0419,
-      "eval_samples_per_second": 17.592,
-      "eval_steps_per_second": 2.202,
-      "eval_wer": 0.21958757093892228,
+      "epoch": 23.99,
+      "eval_loss": 0.33075791597366333,
+      "eval_runtime": 205.0698,
+      "eval_samples_per_second": 17.531,
+      "eval_steps_per_second": 2.194,
+      "eval_wer": 0.2171437865972674,
       "step": 3000
     },
     {
-      "epoch": 25.41,
-      "learning_rate": 9.753978671041838e-05,
-      "loss": 0.0189,
+      "epoch": 24.79,
+      "learning_rate": 9.759887910328262e-05,
+      "loss": 0.0213,
       "step": 3100
     },
     {
-      "epoch": 25.41,
-      "eval_loss": 0.3368154764175415,
-      "eval_runtime": 203.168,
-      "eval_samples_per_second": 17.148,
-      "eval_steps_per_second": 2.146,
-      "eval_wer": 0.21958757093892228,
+      "epoch": 24.79,
+      "eval_loss": 0.3419627845287323,
+      "eval_runtime": 206.2312,
+      "eval_samples_per_second": 17.432,
+      "eval_steps_per_second": 2.182,
+      "eval_wer": 0.21936673172847537,
       "step": 3100
     },
     {
-      "epoch": 26.23,
-      "learning_rate": 9.74577522559475e-05,
-      "loss": 0.0176,
+      "epoch": 25.6,
+      "learning_rate": 9.751881505204163e-05,
+      "loss": 0.0212,
       "step": 3200
     },
     {
-      "epoch": 26.23,
-      "eval_loss": 0.34034982323646545,
-      "eval_runtime": 195.3083,
-      "eval_samples_per_second": 17.838,
-      "eval_steps_per_second": 2.232,
-      "eval_wer": 0.2207675450918694,
+      "epoch": 25.6,
+      "eval_loss": 0.3389272093772888,
+      "eval_runtime": 205.5665,
+      "eval_samples_per_second": 17.488,
+      "eval_steps_per_second": 2.189,
+      "eval_wer": 0.21519193233571893,
       "step": 3200
     },
     {
-      "epoch": 27.05,
-      "learning_rate": 9.737571780147663e-05,
-      "loss": 0.0191,
+      "epoch": 26.4,
+      "learning_rate": 9.743875100080065e-05,
+      "loss": 0.0208,
       "step": 3300
     },
     {
-      "epoch": 27.05,
-      "eval_loss": 0.3415772616863251,
-      "eval_runtime": 201.8077,
-      "eval_samples_per_second": 17.264,
-      "eval_steps_per_second": 2.16,
-      "eval_wer": 0.2209361128280047,
+      "epoch": 26.4,
+      "eval_loss": 0.32781022787094116,
+      "eval_runtime": 208.0862,
+      "eval_samples_per_second": 17.276,
+      "eval_steps_per_second": 2.163,
+      "eval_wer": 0.21275211450878334,
       "step": 3300
     },
     {
-      "epoch": 27.87,
-      "learning_rate": 9.729368334700575e-05,
-      "loss": 0.0203,
+      "epoch": 27.2,
+      "learning_rate": 9.735868694955965e-05,
+      "loss": 0.0212,
       "step": 3400
     },
     {
-      "epoch": 27.87,
-      "eval_loss": 0.3269742429256439,
-      "eval_runtime": 198.0471,
-      "eval_samples_per_second": 17.592,
-      "eval_steps_per_second": 2.201,
-      "eval_wer": 0.21840759678597516,
+      "epoch": 27.2,
+      "eval_loss": 0.3169388771057129,
+      "eval_runtime": 206.029,
+      "eval_samples_per_second": 17.449,
+      "eval_steps_per_second": 2.184,
+      "eval_wer": 0.21497505963999133,
       "step": 3400
     },
     {
-      "epoch": 28.69,
-      "learning_rate": 9.721164889253487e-05,
-      "loss": 0.0178,
+      "epoch": 27.99,
+      "learning_rate": 9.727862289831866e-05,
+      "loss": 0.0213,
       "step": 3500
     },
     {
-      "epoch": 28.69,
-      "eval_loss": 0.3441150188446045,
-      "eval_runtime": 198.5015,
-      "eval_samples_per_second": 17.552,
-      "eval_steps_per_second": 2.196,
-      "eval_wer": 0.22155419452716751,
+      "epoch": 27.99,
+      "eval_loss": 0.31786179542541504,
+      "eval_runtime": 198.632,
+      "eval_samples_per_second": 18.099,
+      "eval_steps_per_second": 2.265,
+      "eval_wer": 0.21757753198872262,
       "step": 3500
     },
     {
-      "epoch": 29.51,
-      "learning_rate": 9.712961443806398e-05,
-      "loss": 0.0178,
+      "epoch": 28.79,
+      "learning_rate": 9.719855884707766e-05,
+      "loss": 0.0213,
       "step": 3600
     },
     {
-      "epoch": 29.51,
-      "eval_loss": 0.3359436094760895,
-      "eval_runtime": 196.1391,
-      "eval_samples_per_second": 17.763,
-      "eval_steps_per_second": 2.223,
-      "eval_wer": 0.22172276226330281,
+      "epoch": 28.79,
+      "eval_loss": 0.32053136825561523,
+      "eval_runtime": 201.1109,
+      "eval_samples_per_second": 17.876,
+      "eval_steps_per_second": 2.238,
+      "eval_wer": 0.2203426588592496,
       "step": 3600
     },
     {
-      "epoch": 30.33,
-      "learning_rate": 9.704757998359312e-05,
-      "loss": 0.0202,
+      "epoch": 29.6,
+      "learning_rate": 9.711849479583668e-05,
+      "loss": 0.0217,
       "step": 3700
     },
     {
-      "epoch": 30.33,
-      "eval_loss": 0.3397650122642517,
-      "eval_runtime": 196.1562,
-      "eval_samples_per_second": 17.761,
-      "eval_steps_per_second": 2.223,
-      "eval_wer": 0.22503792774063044,
+      "epoch": 29.6,
+      "eval_loss": 0.33174052834510803,
+      "eval_runtime": 191.0157,
+      "eval_samples_per_second": 18.82,
+      "eval_steps_per_second": 2.356,
+      "eval_wer": 0.2168726957276079,
       "step": 3700
     },
     {
-      "epoch": 31.15,
-      "learning_rate": 9.696554552912223e-05,
-      "loss": 0.0203,
+      "epoch": 30.4,
+      "learning_rate": 9.703843074459567e-05,
+      "loss": 0.0221,
       "step": 3800
     },
     {
-      "epoch": 31.15,
-      "eval_loss": 0.32521852850914,
-      "eval_runtime": 197.1456,
-      "eval_samples_per_second": 17.672,
-      "eval_steps_per_second": 2.212,
-      "eval_wer": 0.22571219868517164,
+      "epoch": 30.4,
+      "eval_loss": 0.3100583255290985,
+      "eval_runtime": 190.6873,
+      "eval_samples_per_second": 18.853,
+      "eval_steps_per_second": 2.36,
+      "eval_wer": 0.21774018651051832,
       "step": 3800
     },
     {
-      "epoch": 31.97,
-      "learning_rate": 9.688351107465135e-05,
-      "loss": 0.0221,
+      "epoch": 31.2,
+      "learning_rate": 9.695836669335468e-05,
+      "loss": 0.021,
       "step": 3900
     },
     {
-      "epoch": 31.97,
-      "eval_loss": 0.3298187553882599,
-      "eval_runtime": 198.9372,
-      "eval_samples_per_second": 17.513,
-      "eval_steps_per_second": 2.192,
-      "eval_wer": 0.22728549755576782,
+      "epoch": 31.2,
+      "eval_loss": 0.33132240176200867,
+      "eval_runtime": 187.5625,
+      "eval_samples_per_second": 19.167,
+      "eval_steps_per_second": 2.399,
+      "eval_wer": 0.21741487746692692,
       "step": 3900
     },
     {
-      "epoch": 32.79,
-      "learning_rate": 9.680147662018048e-05,
-      "loss": 0.0206,
+      "epoch": 31.99,
+      "learning_rate": 9.68783026421137e-05,
+      "loss": 0.0212,
       "step": 4000
     },
     {
-      "epoch": 32.79,
-      "eval_loss": 0.3412954807281494,
-      "eval_runtime": 195.7736,
-      "eval_samples_per_second": 17.796,
-      "eval_steps_per_second": 2.227,
-      "eval_wer": 0.22363319660616957,
+      "epoch": 31.99,
+      "eval_loss": 0.32063281536102295,
+      "eval_runtime": 192.7124,
+      "eval_samples_per_second": 18.655,
+      "eval_steps_per_second": 2.335,
+      "eval_wer": 0.21340273259596618,
       "step": 4000
-    },
-    {
-      "epoch": 33.61,
-      "learning_rate": 9.671944216570961e-05,
-      "loss": 0.0207,
-      "step": 4100
-    },
-    {
-      "epoch": 33.61,
-      "eval_loss": 0.3309689164161682,
-      "eval_runtime": 201.1531,
-      "eval_samples_per_second": 17.32,
-      "eval_steps_per_second": 2.168,
-      "eval_wer": 0.2248131707591167,
-      "step": 4100
-    },
-    {
-      "epoch": 34.43,
-      "learning_rate": 9.663740771123873e-05,
-      "loss": 0.018,
-      "step": 4200
-    },
-    {
-      "epoch": 34.43,
-      "eval_loss": 0.34882888197898865,
-      "eval_runtime": 197.3922,
-      "eval_samples_per_second": 17.65,
-      "eval_steps_per_second": 2.209,
-      "eval_wer": 0.22329606113389897,
-      "step": 4200
-    },
-    {
-      "epoch": 35.25,
-      "learning_rate": 9.655537325676785e-05,
-      "loss": 0.0193,
-      "step": 4300
-    },
-    {
-      "epoch": 35.25,
-      "eval_loss": 0.30975356698036194,
-      "eval_runtime": 183.1875,
-      "eval_samples_per_second": 19.019,
-      "eval_steps_per_second": 2.38,
-      "eval_wer": 0.22357700736079114,
-      "step": 4300
-    },
-    {
-      "epoch": 36.07,
-      "learning_rate": 9.647333880229696e-05,
-      "loss": 0.0193,
-      "step": 4400
-    },
-    {
-      "epoch": 36.07,
-      "eval_loss": 0.33493199944496155,
-      "eval_runtime": 183.8125,
-      "eval_samples_per_second": 18.954,
-      "eval_steps_per_second": 2.372,
-      "eval_wer": 0.2226779794347362,
-      "step": 4400
-    },
-    {
-      "epoch": 36.89,
-      "learning_rate": 9.63913043478261e-05,
-      "loss": 0.0179,
-      "step": 4500
-    },
-    {
-      "epoch": 36.89,
-      "eval_loss": 0.339138388633728,
-      "eval_runtime": 183.2812,
-      "eval_samples_per_second": 19.009,
-      "eval_steps_per_second": 2.379,
-      "eval_wer": 0.2220598977355734,
-      "step": 4500
-    },
-    {
-      "epoch": 37.7,
-      "learning_rate": 9.630926989335521e-05,
-      "loss": 0.0188,
-      "step": 4600
-    },
-    {
-      "epoch": 37.7,
-      "eval_loss": 0.3318282663822174,
-      "eval_runtime": 186.1719,
-      "eval_samples_per_second": 18.714,
-      "eval_steps_per_second": 2.342,
-      "eval_wer": 0.22408271056919707,
-      "step": 4600
-    },
-    {
-      "epoch": 38.52,
-      "learning_rate": 9.622805578342905e-05,
-      "loss": 0.0191,
-      "step": 4700
-    },
-    {
-      "epoch": 38.52,
-      "eval_loss": 0.3265712261199951,
-      "eval_runtime": 184.3594,
-      "eval_samples_per_second": 18.898,
-      "eval_steps_per_second": 2.365,
-      "eval_wer": 0.2248131707591167,
-      "step": 4700
-    },
-    {
-      "epoch": 39.34,
-      "learning_rate": 9.614602132895817e-05,
-      "loss": 0.02,
-      "step": 4800
-    },
-    {
-      "epoch": 39.34,
-      "eval_loss": 0.32977160811424255,
-      "eval_runtime": 183.375,
-      "eval_samples_per_second": 18.999,
-      "eval_steps_per_second": 2.378,
-      "eval_wer": 0.22335225037927742,
-      "step": 4800
-    },
-    {
-      "epoch": 40.16,
-      "learning_rate": 9.606398687448729e-05,
-      "loss": 0.0194,
-      "step": 4900
-    },
-    {
-      "epoch": 40.16,
-      "eval_loss": 0.33653536438941956,
-      "eval_runtime": 182.625,
-      "eval_samples_per_second": 19.077,
-      "eval_steps_per_second": 2.387,
-      "eval_wer": 0.22520649547676574,
-      "step": 4900
-    },
-    {
-      "epoch": 40.98,
-      "learning_rate": 9.59819524200164e-05,
-      "loss": 0.019,
-      "step": 5000
-    },
-    {
-      "epoch": 40.98,
-      "eval_loss": 0.3455631732940674,
-      "eval_runtime": 185.0148,
-      "eval_samples_per_second": 18.831,
-      "eval_steps_per_second": 2.357,
-      "eval_wer": 0.21998089565657133,
-      "step": 5000
-    },
-    {
-      "epoch": 41.8,
-      "learning_rate": 9.589991796554553e-05,
-      "loss": 0.026,
-      "step": 5100
-    },
-    {
-      "epoch": 41.8,
-      "eval_loss": 0.31202027201652527,
-      "eval_runtime": 187.5,
-      "eval_samples_per_second": 18.581,
-      "eval_steps_per_second": 2.325,
-      "eval_wer": 0.21953138169354386,
-      "step": 5100
-    },
-    {
-      "epoch": 42.62,
-      "learning_rate": 9.581788351107465e-05,
-      "loss": 0.0247,
-      "step": 5200
-    },
-    {
-      "epoch": 42.62,
-      "eval_loss": 0.3445728123188019,
-      "eval_runtime": 183.4063,
-      "eval_samples_per_second": 18.996,
-      "eval_steps_per_second": 2.377,
-      "eval_wer": 0.22082373433724786,
-      "step": 5200
-    },
-    {
-      "epoch": 43.44,
-      "learning_rate": 9.573584905660378e-05,
-      "loss": 0.024,
-      "step": 5300
-    },
-    {
-      "epoch": 43.44,
-      "eval_loss": 0.3246464431285858,
-      "eval_runtime": 197.7008,
-      "eval_samples_per_second": 17.623,
-      "eval_steps_per_second": 2.205,
-      "eval_wer": 0.2205989773557341,
-      "step": 5300
-    },
-    {
-      "epoch": 44.26,
-      "learning_rate": 9.56538146021329e-05,
-      "loss": 0.0221,
-      "step": 5400
-    },
-    {
-      "epoch": 44.26,
-      "eval_loss": 0.3357507586479187,
-      "eval_runtime": 200.2422,
-      "eval_samples_per_second": 17.399,
-      "eval_steps_per_second": 2.177,
-      "eval_wer": 0.21981232792043603,
-      "step": 5400
-    },
-    {
-      "epoch": 45.08,
-      "learning_rate": 9.557178014766203e-05,
-      "loss": 0.023,
-      "step": 5500
-    },
-    {
-      "epoch": 45.08,
-      "eval_loss": 0.33350640535354614,
-      "eval_runtime": 200.3914,
-      "eval_samples_per_second": 17.386,
-      "eval_steps_per_second": 2.176,
-      "eval_wer": 0.22571219868517164,
-      "step": 5500
-    },
-    {
-      "epoch": 45.9,
-      "learning_rate": 9.548974569319115e-05,
-      "loss": 0.0239,
-      "step": 5600
-    },
-    {
-      "epoch": 45.9,
-      "eval_loss": 0.32670244574546814,
-      "eval_runtime": 195.9769,
-      "eval_samples_per_second": 17.778,
-      "eval_steps_per_second": 2.225,
-      "eval_wer": 0.22020565263808506,
-      "step": 5600
-    },
-    {
-      "epoch": 46.72,
-      "learning_rate": 9.540771123872027e-05,
-      "loss": 0.023,
-      "step": 5700
-    },
-    {
-      "epoch": 46.72,
-      "eval_loss": 0.33632007241249084,
-      "eval_runtime": 198.7025,
-      "eval_samples_per_second": 17.534,
-      "eval_steps_per_second": 2.194,
-      "eval_wer": 0.2231836826431421,
-      "step": 5700
-    },
-    {
-      "epoch": 47.54,
-      "learning_rate": 9.532567678424938e-05,
-      "loss": 0.0237,
-      "step": 5800
-    },
-    {
-      "epoch": 47.54,
-      "eval_loss": 0.35056358575820923,
-      "eval_runtime": 199.1109,
-      "eval_samples_per_second": 17.498,
-      "eval_steps_per_second": 2.19,
-      "eval_wer": 0.22099230207338316,
-      "step": 5800
-    },
-    {
-      "epoch": 48.36,
-      "learning_rate": 9.524364232977851e-05,
-      "loss": 0.0251,
-      "step": 5900
-    },
-    {
-      "epoch": 48.36,
-      "eval_loss": 0.33351680636405945,
-      "eval_runtime": 198.4757,
-      "eval_samples_per_second": 17.554,
-      "eval_steps_per_second": 2.197,
-      "eval_wer": 0.22082373433724786,
-      "step": 5900
-    },
-    {
-      "epoch": 49.18,
-      "learning_rate": 9.516160787530763e-05,
-      "loss": 0.024,
-      "step": 6000
-    },
-    {
-      "epoch": 49.18,
-      "eval_loss": 0.3367384672164917,
-      "eval_runtime": 203.7806,
-      "eval_samples_per_second": 17.097,
-      "eval_steps_per_second": 2.14,
-      "eval_wer": 0.2140810248918357,
-      "step": 6000
-    },
-    {
-      "epoch": 50.0,
-      "learning_rate": 9.507957342083675e-05,
-      "loss": 0.023,
-      "step": 6100
-    },
-    {
-      "epoch": 50.0,
-      "eval_loss": 0.33463725447654724,
-      "eval_runtime": 202.5214,
-      "eval_samples_per_second": 17.203,
-      "eval_steps_per_second": 2.153,
-      "eval_wer": 0.2207675450918694,
-      "step": 6100
-    },
-    {
-      "epoch": 50.82,
-      "learning_rate": 9.499753896636588e-05,
-      "loss": 0.0237,
-      "step": 6200
-    },
-    {
-      "epoch": 50.82,
-      "eval_loss": 0.3415164053440094,
-      "eval_runtime": 199.1709,
-      "eval_samples_per_second": 17.493,
-      "eval_steps_per_second": 2.189,
-      "eval_wer": 0.22290273641624994,
-      "step": 6200
-    },
-    {
-      "epoch": 51.64,
-      "learning_rate": 9.4915504511895e-05,
-      "loss": 0.0238,
-      "step": 6300
-    },
-    {
-      "epoch": 51.64,
-      "eval_loss": 0.34561434388160706,
-      "eval_runtime": 196.5595,
-      "eval_samples_per_second": 17.725,
-      "eval_steps_per_second": 2.218,
-      "eval_wer": 0.21638478395235153,
-      "step": 6300
-    },
-    {
-      "epoch": 52.46,
-      "learning_rate": 9.483347005742413e-05,
-      "loss": 0.0242,
-      "step": 6400
-    },
-    {
-      "epoch": 52.46,
-      "eval_loss": 0.34634143114089966,
-      "eval_runtime": 202.0785,
-      "eval_samples_per_second": 17.241,
-      "eval_steps_per_second": 2.158,
-      "eval_wer": 0.22666741585660505,
-      "step": 6400
-    },
-    {
-      "epoch": 53.28,
-      "learning_rate": 9.475143560295325e-05,
-      "loss": 0.0238,
-      "step": 6500
-    },
-    {
-      "epoch": 53.28,
-      "eval_loss": 0.3443451225757599,
-      "eval_runtime": 198.1661,
-      "eval_samples_per_second": 17.581,
-      "eval_steps_per_second": 2.2,
-      "eval_wer": 0.21722762263302803,
-      "step": 6500
-    },
-    {
-      "epoch": 54.1,
-      "learning_rate": 9.466940114848236e-05,
-      "loss": 0.0247,
-      "step": 6600
-    },
-    {
-      "epoch": 54.1,
-      "eval_loss": 0.33527693152427673,
-      "eval_runtime": 202.0925,
-      "eval_samples_per_second": 17.24,
-      "eval_steps_per_second": 2.157,
-      "eval_wer": 0.21941900320278698,
-      "step": 6600
-    },
-    {
-      "epoch": 54.92,
-      "learning_rate": 9.45873666940115e-05,
-      "loss": 0.024,
-      "step": 6700
-    },
-    {
-      "epoch": 54.92,
-      "eval_loss": 0.33691638708114624,
-      "eval_runtime": 203.115,
-      "eval_samples_per_second": 17.153,
-      "eval_steps_per_second": 2.147,
-      "eval_wer": 0.21481148508175535,
-      "step": 6700
-    },
-    {
-      "epoch": 55.74,
-      "learning_rate": 9.450533223954061e-05,
-      "loss": 0.023,
-      "step": 6800
-    },
-    {
-      "epoch": 55.74,
-      "eval_loss": 0.3394332826137543,
-      "eval_runtime": 205.3023,
-      "eval_samples_per_second": 16.97,
-      "eval_steps_per_second": 2.124,
-      "eval_wer": 0.21807046131370456,
-      "step": 6800
-    },
-    {
-      "epoch": 56.56,
-      "learning_rate": 9.442329778506973e-05,
-      "loss": 0.0226,
-      "step": 6900
-    },
-    {
-      "epoch": 56.56,
-      "eval_loss": 0.3425767123699188,
-      "eval_runtime": 204.2505,
-      "eval_samples_per_second": 17.057,
-      "eval_steps_per_second": 2.135,
-      "eval_wer": 0.21818283980446143,
-      "step": 6900
-    },
-    {
-      "epoch": 57.38,
-      "learning_rate": 9.434126333059886e-05,
-      "loss": 0.0233,
-      "step": 7000
-    },
-    {
-      "epoch": 57.38,
-      "eval_loss": 0.3361387252807617,
-      "eval_runtime": 204.131,
-      "eval_samples_per_second": 17.067,
-      "eval_steps_per_second": 2.136,
-      "eval_wer": 0.21981232792043603,
-      "step": 7000
-    },
-    {
-      "epoch": 58.2,
-      "learning_rate": 9.425922887612798e-05,
-      "loss": 0.0231,
-      "step": 7100
-    },
-    {
-      "epoch": 58.2,
-      "eval_loss": 0.3401791751384735,
-      "eval_runtime": 204.9298,
-      "eval_samples_per_second": 17.001,
-      "eval_steps_per_second": 2.128,
-      "eval_wer": 0.2207675450918694,
-      "step": 7100
-    },
-    {
-      "epoch": 59.02,
-      "learning_rate": 9.41780147662018e-05,
-      "loss": 0.0231,
-      "step": 7200
-    },
-    {
-      "epoch": 59.02,
-      "eval_loss": 0.32741424441337585,
-      "eval_runtime": 205.0196,
-      "eval_samples_per_second": 16.993,
-      "eval_steps_per_second": 2.127,
-      "eval_wer": 0.22295892566162837,
-      "step": 7200
-    },
-    {
-      "epoch": 59.84,
-      "learning_rate": 9.409598031173093e-05,
-      "loss": 0.0242,
-      "step": 7300
-    },
-    {
-      "epoch": 59.84,
-      "eval_loss": 0.3416125774383545,
-      "eval_runtime": 205.8312,
-      "eval_samples_per_second": 16.926,
-      "eval_steps_per_second": 2.118,
-      "eval_wer": 0.21773332584143396,
-      "step": 7300
-    },
-    {
-      "epoch": 60.66,
-      "learning_rate": 9.401394585726005e-05,
-      "loss": 0.0221,
-      "step": 7400
-    },
-    {
-      "epoch": 60.66,
-      "eval_loss": 0.3495969772338867,
-      "eval_runtime": 207.6167,
-      "eval_samples_per_second": 16.781,
-      "eval_steps_per_second": 2.1,
-      "eval_wer": 0.21874473225824578,
-      "step": 7400
-    },
-    {
-      "epoch": 61.48,
-      "learning_rate": 9.393191140278917e-05,
-      "loss": 0.0231,
-      "step": 7500
-    },
-    {
-      "epoch": 61.48,
-      "eval_loss": 0.33186405897140503,
-      "eval_runtime": 206.7658,
-      "eval_samples_per_second": 16.85,
-      "eval_steps_per_second": 2.109,
-      "eval_wer": 0.22155419452716751,
-      "step": 7500
-    },
-    {
-      "epoch": 62.3,
-      "learning_rate": 9.38498769483183e-05,
-      "loss": 0.021,
-      "step": 7600
-    },
-    {
-      "epoch": 62.3,
-      "eval_loss": 0.3377366065979004,
-      "eval_runtime": 203.3638,
-      "eval_samples_per_second": 17.132,
-      "eval_steps_per_second": 2.144,
-      "eval_wer": 0.21998089565657133,
-      "step": 7600
-    },
-    {
-      "epoch": 63.11,
-      "learning_rate": 9.376784249384743e-05,
-      "loss": 0.022,
-      "step": 7700
-    },
-    {
-      "epoch": 63.11,
-      "eval_loss": 0.3331567943096161,
-      "eval_runtime": 203.4814,
-      "eval_samples_per_second": 17.122,
-      "eval_steps_per_second": 2.143,
-      "eval_wer": 0.22127324830027534,
-      "step": 7700
-    },
-    {
-      "epoch": 63.93,
-      "learning_rate": 9.368580803937655e-05,
-      "loss": 0.0216,
-      "step": 7800
-    },
-    {
-      "epoch": 63.93,
-      "eval_loss": 0.32978036999702454,
-      "eval_runtime": 204.6963,
-      "eval_samples_per_second": 17.02,
-      "eval_steps_per_second": 2.13,
-      "eval_wer": 0.2230151149070068,
-      "step": 7800
-    },
-    {
-      "epoch": 64.75,
-      "learning_rate": 9.360377358490567e-05,
-      "loss": 0.021,
-      "step": 7900
-    },
-    {
-      "epoch": 64.75,
-      "eval_loss": 0.3439036011695862,
-      "eval_runtime": 203.7134,
-      "eval_samples_per_second": 17.102,
-      "eval_steps_per_second": 2.14,
-      "eval_wer": 0.22329606113389897,
-      "step": 7900
-    },
-    {
-      "epoch": 65.57,
-      "learning_rate": 9.352173913043478e-05,
-      "loss": 0.0217,
-      "step": 8000
-    },
-    {
-      "epoch": 65.57,
-      "eval_loss": 0.34148770570755005,
-      "eval_runtime": 202.6414,
-      "eval_samples_per_second": 17.193,
-      "eval_steps_per_second": 2.152,
-      "eval_wer": 0.21969994942967916,
-      "step": 8000
-    },
-    {
-      "epoch": 66.39,
-      "learning_rate": 9.343970467596391e-05,
-      "loss": 0.021,
-      "step": 8100
-    },
-    {
-      "epoch": 66.39,
-      "eval_loss": 0.3486896753311157,
-      "eval_runtime": 202.1874,
-      "eval_samples_per_second": 17.232,
-      "eval_steps_per_second": 2.156,
-      "eval_wer": 0.21969994942967916,
-      "step": 8100
-    },
-    {
-      "epoch": 67.21,
-      "learning_rate": 9.335767022149303e-05,
-      "loss": 0.0207,
-      "step": 8200
-    },
-    {
-      "epoch": 67.21,
-      "eval_loss": 0.3460260331630707,
-      "eval_runtime": 200.0049,
-      "eval_samples_per_second": 17.42,
-      "eval_steps_per_second": 2.18,
-      "eval_wer": 0.21795808282294768,
-      "step": 8200
-    },
-    {
-      "epoch": 68.03,
-      "learning_rate": 9.327563576702215e-05,
-      "loss": 0.0212,
-      "step": 8300
-    },
-    {
-      "epoch": 68.03,
-      "eval_loss": 0.35000941157341003,
-      "eval_runtime": 202.0373,
-      "eval_samples_per_second": 17.244,
-      "eval_steps_per_second": 2.158,
-      "eval_wer": 0.21627240546159465,
-      "step": 8300
-    },
-    {
-      "epoch": 68.85,
-      "learning_rate": 9.319360131255127e-05,
-      "loss": 0.0223,
-      "step": 8400
-    },
-    {
-      "epoch": 68.85,
-      "eval_loss": 0.33812931180000305,
-      "eval_runtime": 201.8642,
-      "eval_samples_per_second": 17.259,
-      "eval_steps_per_second": 2.16,
-      "eval_wer": 0.2200932741473282,
-      "step": 8400
-    },
-    {
-      "epoch": 69.67,
-      "learning_rate": 9.31123872026251e-05,
-      "loss": 0.0224,
-      "step": 8500
-    },
-    {
-      "epoch": 69.67,
-      "eval_loss": 0.3443692624568939,
-      "eval_runtime": 202.0712,
-      "eval_samples_per_second": 17.241,
-      "eval_steps_per_second": 2.158,
-      "eval_wer": 0.21936281395740856,
-      "step": 8500
-    },
-    {
-      "epoch": 70.49,
-      "learning_rate": 9.303035274815422e-05,
-      "loss": 0.0229,
-      "step": 8600
-    },
-    {
-      "epoch": 70.49,
-      "eval_loss": 0.3441421389579773,
-      "eval_runtime": 203.724,
-      "eval_samples_per_second": 17.102,
-      "eval_steps_per_second": 2.14,
-      "eval_wer": 0.2227341686801146,
-      "step": 8600
-    },
-    {
-      "epoch": 71.31,
-      "learning_rate": 9.294831829368335e-05,
-      "loss": 0.0212,
-      "step": 8700
-    },
-    {
-      "epoch": 71.31,
-      "eval_loss": 0.3485627770423889,
-      "eval_runtime": 202.8979,
-      "eval_samples_per_second": 17.171,
-      "eval_steps_per_second": 2.149,
-      "eval_wer": 0.2218913299994381,
-      "step": 8700
-    },
-    {
-      "epoch": 72.13,
-      "learning_rate": 9.286628383921247e-05,
-      "loss": 0.0221,
-      "step": 8800
-    },
-    {
-      "epoch": 72.13,
-      "eval_loss": 0.33275967836380005,
-      "eval_runtime": 204.5042,
-      "eval_samples_per_second": 17.036,
-      "eval_steps_per_second": 2.132,
-      "eval_wer": 0.22116086980951846,
-      "step": 8800
-    },
-    {
-      "epoch": 72.95,
-      "learning_rate": 9.27842493847416e-05,
-      "loss": 0.02,
-      "step": 8900
-    },
-    {
-      "epoch": 72.95,
-      "eval_loss": 0.3454422950744629,
-      "eval_runtime": 203.7032,
-      "eval_samples_per_second": 17.103,
-      "eval_steps_per_second": 2.14,
-      "eval_wer": 0.2212170590548969,
-      "step": 8900
-    },
-    {
-      "epoch": 73.77,
-      "learning_rate": 9.270221493027072e-05,
-      "loss": 0.0202,
-      "step": 9000
-    },
-    {
-      "epoch": 73.77,
-      "eval_loss": 0.36007246375083923,
-      "eval_runtime": 200.6262,
-      "eval_samples_per_second": 17.366,
-      "eval_steps_per_second": 2.173,
-      "eval_wer": 0.21677810867000055,
-      "step": 9000
-    },
-    {
-      "epoch": 74.59,
-      "learning_rate": 9.262018047579985e-05,
-      "loss": 0.021,
-      "step": 9100
-    },
-    {
-      "epoch": 74.59,
-      "eval_loss": 0.34157735109329224,
-      "eval_runtime": 204.0017,
-      "eval_samples_per_second": 17.078,
-      "eval_steps_per_second": 2.137,
-      "eval_wer": 0.2205427881103557,
-      "step": 9100
-    },
-    {
-      "epoch": 75.41,
-      "learning_rate": 9.253814602132897e-05,
-      "loss": 0.0217,
-      "step": 9200
-    },
-    {
-      "epoch": 75.41,
-      "eval_loss": 0.35333356261253357,
-      "eval_runtime": 207.6876,
-      "eval_samples_per_second": 16.775,
-      "eval_steps_per_second": 2.099,
-      "eval_wer": 0.2230151149070068,
-      "step": 9200
-    },
-    {
-      "epoch": 76.23,
-      "learning_rate": 9.245611156685808e-05,
-      "loss": 0.0209,
-      "step": 9300
-    },
-    {
-      "epoch": 76.23,
-      "eval_loss": 0.35574042797088623,
-      "eval_runtime": 204.8643,
-      "eval_samples_per_second": 17.006,
-      "eval_steps_per_second": 2.128,
-      "eval_wer": 0.22020565263808506,
-      "step": 9300
-    },
-    {
-      "epoch": 77.05,
-      "learning_rate": 9.23740771123872e-05,
-      "loss": 0.0214,
-      "step": 9400
-    },
-    {
-      "epoch": 77.05,
-      "eval_loss": 0.3214012086391449,
-      "eval_runtime": 203.9509,
-      "eval_samples_per_second": 17.083,
-      "eval_steps_per_second": 2.138,
-      "eval_wer": 0.21851997527673203,
-      "step": 9400
-    },
-    {
-      "epoch": 77.87,
-      "learning_rate": 9.229204265791633e-05,
-      "loss": 0.0212,
-      "step": 9500
-    },
-    {
-      "epoch": 77.87,
-      "eval_loss": 0.33761337399482727,
-      "eval_runtime": 205.1801,
-      "eval_samples_per_second": 16.98,
-      "eval_steps_per_second": 2.125,
-      "eval_wer": 0.21745237961454178,
-      "step": 9500
-    },
-    {
-      "epoch": 78.69,
-      "learning_rate": 9.221000820344545e-05,
-      "loss": 0.0214,
-      "step": 9600
-    },
-    {
-      "epoch": 78.69,
-      "eval_loss": 0.32581719756126404,
-      "eval_runtime": 204.8134,
-      "eval_samples_per_second": 17.011,
-      "eval_steps_per_second": 2.129,
-      "eval_wer": 0.21722762263302803,
-      "step": 9600
-    },
-    {
-      "epoch": 79.51,
-      "learning_rate": 9.212797374897457e-05,
-      "loss": 0.022,
-      "step": 9700
-    },
-    {
-      "epoch": 79.51,
-      "eval_loss": 0.3441867530345917,
-      "eval_runtime": 207.7734,
-      "eval_samples_per_second": 16.768,
-      "eval_steps_per_second": 2.098,
-      "eval_wer": 0.2220598977355734,
-      "step": 9700
-    },
-    {
-      "epoch": 80.33,
-      "learning_rate": 9.204593929450369e-05,
-      "loss": 0.0236,
-      "step": 9800
-    },
-    {
-      "epoch": 80.33,
-      "eval_loss": 0.33785006403923035,
-      "eval_runtime": 206.8523,
-      "eval_samples_per_second": 16.843,
-      "eval_steps_per_second": 2.108,
-      "eval_wer": 0.22458841377760297,
-      "step": 9800
-    },
-    {
-      "epoch": 81.15,
-      "learning_rate": 9.196390484003282e-05,
-      "loss": 0.0231,
-      "step": 9900
-    },
-    {
-      "epoch": 81.15,
-      "eval_loss": 0.33509162068367004,
-      "eval_runtime": 205.9688,
-      "eval_samples_per_second": 16.915,
-      "eval_steps_per_second": 2.117,
-      "eval_wer": 0.2182952182952183,
-      "step": 9900
-    },
-    {
-      "epoch": 81.97,
-      "learning_rate": 9.188187038556195e-05,
-      "loss": 0.0206,
-      "step": 10000
-    },
-    {
-      "epoch": 81.97,
-      "eval_loss": 0.331687331199646,
-      "eval_runtime": 207.4393,
-      "eval_samples_per_second": 16.795,
-      "eval_steps_per_second": 2.102,
-      "eval_wer": 0.21644097319772995,
-      "step": 10000
-    },
-    {
-      "epoch": 82.79,
-      "learning_rate": 9.179983593109106e-05,
-      "loss": 0.0199,
-      "step": 10100
-    },
-    {
-      "epoch": 82.79,
-      "eval_loss": 0.3436143398284912,
-      "eval_runtime": 207.0841,
-      "eval_samples_per_second": 16.824,
-      "eval_steps_per_second": 2.105,
-      "eval_wer": 0.2175085688599202,
-      "step": 10100
-    },
-    {
-      "epoch": 83.61,
-      "learning_rate": 9.171780147662018e-05,
-      "loss": 0.0201,
-      "step": 10200
-    },
-    {
-      "epoch": 83.61,
-      "eval_loss": 0.3532947599887848,
-      "eval_runtime": 206.5219,
-      "eval_samples_per_second": 16.87,
-      "eval_steps_per_second": 2.111,
-      "eval_wer": 0.21689048716075743,
-      "step": 10200
-    },
-    {
-      "epoch": 84.43,
-      "learning_rate": 9.163576702214931e-05,
-      "loss": 0.0194,
-      "step": 10300
-    },
-    {
-      "epoch": 84.43,
-      "eval_loss": 0.3359845280647278,
-      "eval_runtime": 206.1506,
-      "eval_samples_per_second": 16.9,
-      "eval_steps_per_second": 2.115,
-      "eval_wer": 0.2166657301792437,
-      "step": 10300
-    },
-    {
-      "epoch": 85.25,
-      "learning_rate": 9.155373256767843e-05,
-      "loss": 0.0195,
-      "step": 10400
-    },
-    {
-      "epoch": 85.25,
-      "eval_loss": 0.34440097212791443,
-      "eval_runtime": 201.6413,
-      "eval_samples_per_second": 17.278,
-      "eval_steps_per_second": 2.162,
-      "eval_wer": 0.21773332584143396,
-      "step": 10400
-    },
-    {
-      "epoch": 86.07,
-      "learning_rate": 9.147169811320755e-05,
-      "loss": 0.0214,
-      "step": 10500
-    },
-    {
-      "epoch": 86.07,
-      "eval_loss": 0.355741411447525,
-      "eval_runtime": 201.504,
-      "eval_samples_per_second": 17.29,
-      "eval_steps_per_second": 2.164,
-      "eval_wer": 0.21756475810529866,
-      "step": 10500
-    },
-    {
-      "epoch": 86.89,
-      "learning_rate": 9.138966365873666e-05,
-      "loss": 0.0201,
-      "step": 10600
-    },
-    {
-      "epoch": 86.89,
-      "eval_loss": 0.3295002281665802,
-      "eval_runtime": 202.835,
-      "eval_samples_per_second": 17.177,
-      "eval_steps_per_second": 2.15,
-      "eval_wer": 0.21812665055908298,
-      "step": 10600
-    },
-    {
-      "epoch": 87.7,
-      "learning_rate": 9.13076292042658e-05,
-      "loss": 0.0201,
-      "step": 10700
-    },
-    {
-      "epoch": 87.7,
-      "eval_loss": 0.32654306292533875,
-      "eval_runtime": 204.4319,
-      "eval_samples_per_second": 17.042,
-      "eval_steps_per_second": 2.133,
-      "eval_wer": 0.21593526998932405,
-      "step": 10700
-    },
-    {
-      "epoch": 88.52,
-      "learning_rate": 9.122559474979491e-05,
-      "loss": 0.0205,
-      "step": 10800
-    },
-    {
-      "epoch": 88.52,
-      "eval_loss": 0.35505411028862,
-      "eval_runtime": 200.1439,
-      "eval_samples_per_second": 17.407,
-      "eval_steps_per_second": 2.178,
-      "eval_wer": 0.22374557509692644,
-      "step": 10800
-    },
-    {
-      "epoch": 89.34,
-      "learning_rate": 9.114356029532404e-05,
-      "loss": 0.0207,
-      "step": 10900
-    },
-    {
-      "epoch": 89.34,
-      "eval_loss": 0.32287225127220154,
-      "eval_runtime": 202.1756,
-      "eval_samples_per_second": 17.233,
-      "eval_steps_per_second": 2.157,
-      "eval_wer": 0.22144181603641064,
-      "step": 10900
-    },
-    {
-      "epoch": 90.16,
-      "learning_rate": 9.106152584085316e-05,
-      "loss": 0.0196,
-      "step": 11000
-    },
-    {
-      "epoch": 90.16,
-      "eval_loss": 0.3513171374797821,
-      "eval_runtime": 200.3912,
-      "eval_samples_per_second": 17.386,
-      "eval_steps_per_second": 2.176,
-      "eval_wer": 0.2175085688599202,
-      "step": 11000
-    },
-    {
-      "epoch": 90.98,
-      "learning_rate": 9.097949138638229e-05,
-      "loss": 0.0192,
-      "step": 11100
-    },
-    {
-      "epoch": 90.98,
-      "eval_loss": 0.35852402448654175,
-      "eval_runtime": 203.5945,
-      "eval_samples_per_second": 17.112,
-      "eval_steps_per_second": 2.142,
-      "eval_wer": 0.21998089565657133,
-      "step": 11100
-    },
-    {
-      "epoch": 91.8,
-      "learning_rate": 9.089745693191141e-05,
-      "loss": 0.0199,
-      "step": 11200
-    },
-    {
-      "epoch": 91.8,
-      "eval_loss": 0.3458695411682129,
-      "eval_runtime": 186.9089,
-      "eval_samples_per_second": 18.64,
-      "eval_steps_per_second": 2.333,
-      "eval_wer": 0.22127324830027534,
-      "step": 11200
-    },
-    {
-      "epoch": 92.62,
-      "learning_rate": 9.081542247744053e-05,
-      "loss": 0.0191,
-      "step": 11300
-    },
-    {
-      "epoch": 92.62,
-      "eval_loss": 0.3403360843658447,
-      "eval_runtime": 186.3002,
-      "eval_samples_per_second": 18.701,
-      "eval_steps_per_second": 2.34,
-      "eval_wer": 0.22172276226330281,
-      "step": 11300
-    },
-    {
-      "epoch": 93.44,
-      "learning_rate": 9.073338802296966e-05,
-      "loss": 0.019,
-      "step": 11400
-    },
-    {
-      "epoch": 93.44,
-      "eval_loss": 0.34108027815818787,
-      "eval_runtime": 184.4375,
-      "eval_samples_per_second": 18.89,
-      "eval_steps_per_second": 2.364,
-      "eval_wer": 0.21486767432713377,
-      "step": 11400
-    },
-    {
-      "epoch": 94.26,
-      "learning_rate": 9.065135356849878e-05,
-      "loss": 0.0197,
-      "step": 11500
-    },
-    {
-      "epoch": 94.26,
-      "eval_loss": 0.3422238528728485,
-      "eval_runtime": 191.25,
-      "eval_samples_per_second": 18.217,
-      "eval_steps_per_second": 2.28,
-      "eval_wer": 0.22082373433724786,
-      "step": 11500
-    },
-    {
-      "epoch": 95.08,
-      "learning_rate": 9.056931911402789e-05,
-      "loss": 0.0197,
-      "step": 11600
-    },
-    {
-      "epoch": 95.08,
-      "eval_loss": 0.33903756737709045,
-      "eval_runtime": 208.792,
-      "eval_samples_per_second": 16.686,
-      "eval_steps_per_second": 2.088,
-      "eval_wer": 0.21986851716581446,
-      "step": 11600
-    },
-    {
-      "epoch": 95.9,
-      "learning_rate": 9.048728465955701e-05,
-      "loss": 0.0207,
-      "step": 11700
-    },
-    {
-      "epoch": 95.9,
-      "eval_loss": 0.322264164686203,
-      "eval_runtime": 207.201,
-      "eval_samples_per_second": 16.815,
-      "eval_steps_per_second": 2.104,
-      "eval_wer": 0.21351913243805135,
-      "step": 11700
     }
   ],
-  "max_steps": 122000,
+  "max_steps": 125000,
   "num_train_epochs": 1000,
-  "total_flos": 2.640848798488764e+20,
+  "total_flos": 9.112861160368682e+19,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f52fa40ebbde5a5ec6af8d2b2cceb3adbc9f404cf3afe5cd25bb4606ecbb4bf0
+oid sha256:a4775b7d54e5ed7fc71d018fd7cb01140ef925d4333446c094a86299d55c0701
 size 3323