End of training

README.md

@@ -1,6 +1,10 @@
 ---
+language:
+- tr
 base_model: ylacombe/w2v-bert-2.0
 tags:
+- automatic-speech-recognition
+- mozilla-foundation/common_voice_16_0
 - generated_from_trainer
 datasets:
 - common_voice_16_0
@@ -13,11 +17,11 @@ model-index:
       name: Automatic Speech Recognition
       type: automatic-speech-recognition
     dataset:
-      name: common_voice_16_0
+      name: MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - TR
       type: common_voice_16_0
       config: tr
       split: test
-      args: tr
+      args: 'Config: tr, Training split: train+validation, Eval split: test'
     metrics:
     - name: Wer
       type: wer
@@ -29,7 +33,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # wav2vec2-common_voice-tr-demo
 
-This model is a fine-tuned version of [ylacombe/w2v-bert-2.0](https://huggingface.co/ylacombe/w2v-bert-2.0) on the common_voice_16_0 dataset.
+This model is a fine-tuned version of [ylacombe/w2v-bert-2.0](https://huggingface.co/ylacombe/w2v-bert-2.0) on the MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - TR dataset.
 It achieves the following results on the evaluation set:
 - Loss: nan
 - Wer: 1.0

all_results.json

@@ -1,14 +1,14 @@
 {
     "epoch": 15.0,
-    "eval_loss": 3.0581979751586914,
-    "eval_runtime": 309.3284,
+    "eval_loss": NaN,
+    "eval_runtime": 324.5516,
     "eval_samples": 11035,
-    "eval_samples_per_second": 35.674,
-    "eval_steps_per_second": 1.115,
+    "eval_samples_per_second": 34.001,
+    "eval_steps_per_second": 1.063,
     "eval_wer": 1.0,
-    "train_loss": 3.3253096312547563,
-    "train_runtime": 44962.3834,
+    "train_loss": 1.129231213434646,
+    "train_runtime": 45647.7738,
     "train_samples": 43798,
-    "train_samples_per_second": 14.612,
-    "train_steps_per_second": 0.365
+    "train_samples_per_second": 14.392,
+    "train_steps_per_second": 0.36
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 15.0,
-    "eval_loss": 3.0581979751586914,
-    "eval_runtime": 309.3284,
+    "eval_loss": NaN,
+    "eval_runtime": 324.5516,
     "eval_samples": 11035,
-    "eval_samples_per_second": 35.674,
-    "eval_steps_per_second": 1.115,
+    "eval_samples_per_second": 34.001,
+    "eval_steps_per_second": 1.063,
     "eval_wer": 1.0
 }

runs/Jan04_20-37-14_vorace/events.out.tfevents.1704447903.vorace.1537045.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09560f79aeeebf2523077eeee9b0ff225d0f827645df60dfe224ba16ac54d122
+size 364

runs/Jan05_08-39-35_vorace/events.out.tfevents.1704445083.vorace.1722965.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d83369742a1f74133bc19aca815f10b018697dbe547b6f55b85b1910cbc89544
-size 6761
+oid sha256:6c84137a9499be9a4be711454736f6dd77efee3a4093c4f6c586261b7c1810bf
+size 6918

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "train_loss": 3.3253096312547563,
-    "train_runtime": 44962.3834,
+    "train_loss": 1.129231213434646,
+    "train_runtime": 45647.7738,
     "train_samples": 43798,
-    "train_samples_per_second": 14.612,
-    "train_steps_per_second": 0.365
+    "train_samples_per_second": 14.392,
+    "train_steps_per_second": 0.36
 }

trainer_state.json

@@ -10,679 +10,679 @@
   "log_history": [
     {
       "epoch": 0.27,
-      "eval_loss": 7.266334533691406,
-      "eval_runtime": 325.0132,
-      "eval_samples_per_second": 33.952,
-      "eval_steps_per_second": 1.061,
+      "eval_loss": 3.293024778366089,
+      "eval_runtime": 329.2436,
+      "eval_samples_per_second": 33.516,
+      "eval_steps_per_second": 1.048,
       "eval_wer": 1.0,
       "step": 300
     },
     {
       "epoch": 0.46,
-      "learning_rate": 7.14420628442998e-05,
-      "loss": 10.5256,
+      "learning_rate": 0.0007389237223065434,
+      "loss": 5.6462,
       "step": 500
     },
     {
       "epoch": 0.55,
-      "eval_loss": 3.0892837047576904,
-      "eval_runtime": 316.7833,
-      "eval_samples_per_second": 34.835,
-      "eval_steps_per_second": 1.089,
+      "eval_loss": 3.415907859802246,
+      "eval_runtime": 328.5335,
+      "eval_samples_per_second": 33.589,
+      "eval_steps_per_second": 1.05,
       "eval_wer": 1.0,
       "step": 600
     },
     {
       "epoch": 0.82,
-      "eval_loss": 3.061225652694702,
-      "eval_runtime": 483.1327,
-      "eval_samples_per_second": 22.841,
-      "eval_steps_per_second": 0.714,
+      "eval_loss": 3.4422173500061035,
+      "eval_runtime": 338.1217,
+      "eval_samples_per_second": 32.636,
+      "eval_steps_per_second": 1.02,
       "eval_wer": 1.0,
       "step": 900
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0001434602713566988,
-      "loss": 2.9795,
+      "learning_rate": 0.0014838065068897525,
+      "loss": 3.3522,
       "step": 1000
     },
     {
       "epoch": 1.1,
-      "eval_loss": 2.9936766624450684,
-      "eval_runtime": 320.3828,
-      "eval_samples_per_second": 34.443,
-      "eval_steps_per_second": 1.077,
+      "eval_loss": 3.3719358444213867,
+      "eval_runtime": 340.7912,
+      "eval_samples_per_second": 32.381,
+      "eval_steps_per_second": 1.012,
       "eval_wer": 1.0,
       "step": 1200
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0002154784798690978,
-      "loss": 2.9564,
+      "learning_rate": 0.002228689291472962,
+      "loss": 3.2605,
       "step": 1500
     },
     {
       "epoch": 1.37,
-      "eval_loss": 3.242413282394409,
-      "eval_runtime": 328.1378,
-      "eval_samples_per_second": 33.629,
-      "eval_steps_per_second": 1.051,
+      "eval_loss": 3.4025769233703613,
+      "eval_runtime": 335.1987,
+      "eval_samples_per_second": 32.921,
+      "eval_steps_per_second": 1.029,
       "eval_wer": 1.0,
       "step": 1500
     },
     {
       "epoch": 1.64,
-      "eval_loss": 3.2866387367248535,
-      "eval_runtime": 325.8189,
-      "eval_samples_per_second": 33.869,
-      "eval_steps_per_second": 1.059,
+      "eval_loss": 3.444835662841797,
+      "eval_runtime": 332.4199,
+      "eval_samples_per_second": 33.196,
+      "eval_steps_per_second": 1.038,
       "eval_wer": 1.0,
       "step": 1800
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0002874966883814968,
-      "loss": 3.1552,
+      "learning_rate": 0.0029735720760561708,
+      "loss": 3.2766,
       "step": 2000
     },
     {
       "epoch": 1.92,
-      "eval_loss": 3.6338589191436768,
-      "eval_runtime": 322.0052,
-      "eval_samples_per_second": 34.27,
-      "eval_steps_per_second": 1.071,
-      "eval_wer": 1.0,
+      "eval_loss": 3.473637104034424,
+      "eval_runtime": 334.5697,
+      "eval_samples_per_second": 32.983,
+      "eval_steps_per_second": 1.031,
+      "eval_wer": 0.9999093792478477,
       "step": 2100
     },
     {
       "epoch": 2.19,
-      "eval_loss": 3.1184866428375244,
-      "eval_runtime": 316.2742,
-      "eval_samples_per_second": 34.891,
-      "eval_steps_per_second": 1.091,
+      "eval_loss": 3.982806444168091,
+      "eval_runtime": 342.0917,
+      "eval_samples_per_second": 32.257,
+      "eval_steps_per_second": 1.009,
       "eval_wer": 1.0,
       "step": 2400
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00035951489689389575,
-      "loss": 3.2079,
+      "learning_rate": 0.0037184548606393796,
+      "loss": 3.2853,
       "step": 2500
     },
     {
       "epoch": 2.47,
-      "eval_loss": 3.183176040649414,
-      "eval_runtime": 316.7437,
-      "eval_samples_per_second": 34.839,
-      "eval_steps_per_second": 1.089,
+      "eval_loss": 3.553187370300293,
+      "eval_runtime": 329.7168,
+      "eval_samples_per_second": 33.468,
+      "eval_steps_per_second": 1.046,
       "eval_wer": 1.0,
       "step": 2700
     },
     {
       "epoch": 2.74,
-      "learning_rate": 0.00043153310540629475,
-      "loss": 3.1275,
+      "learning_rate": 0.004460358114084256,
+      "loss": 3.3389,
       "step": 3000
     },
     {
       "epoch": 2.74,
-      "eval_loss": 3.3952367305755615,
-      "eval_runtime": 319.4373,
-      "eval_samples_per_second": 34.545,
-      "eval_steps_per_second": 1.08,
+      "eval_loss": 3.781858444213867,
+      "eval_runtime": 323.3731,
+      "eval_samples_per_second": 34.125,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 3000
     },
     {
       "epoch": 3.01,
-      "eval_loss": 3.2981579303741455,
-      "eval_runtime": 327.568,
-      "eval_samples_per_second": 33.688,
-      "eval_steps_per_second": 1.053,
+      "eval_loss": 3.2249505519866943,
+      "eval_runtime": 323.988,
+      "eval_samples_per_second": 34.06,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 3300
     },
     {
       "epoch": 3.2,
-      "learning_rate": 0.0005035513139186939,
-      "loss": 3.0987,
+      "learning_rate": 0.005205240898667465,
+      "loss": 3.2186,
       "step": 3500
     },
     {
       "epoch": 3.29,
-      "eval_loss": 3.103595733642578,
-      "eval_runtime": 327.3342,
-      "eval_samples_per_second": 33.712,
-      "eval_steps_per_second": 1.054,
+      "eval_loss": 3.2372846603393555,
+      "eval_runtime": 324.034,
+      "eval_samples_per_second": 34.055,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 3600
     },
     {
       "epoch": 3.56,
-      "eval_loss": 3.1222941875457764,
-      "eval_runtime": 312.8357,
-      "eval_samples_per_second": 35.274,
-      "eval_steps_per_second": 1.103,
+      "eval_loss": 3.2161905765533447,
+      "eval_runtime": 323.7754,
+      "eval_samples_per_second": 34.082,
+      "eval_steps_per_second": 1.066,
       "eval_wer": 1.0,
       "step": 3900
     },
     {
       "epoch": 3.65,
-      "learning_rate": 0.0005755695224310928,
-      "loss": 2.9301,
+      "learning_rate": 0.005950123683250674,
+      "loss": 3.1916,
       "step": 4000
     },
     {
       "epoch": 3.84,
-      "eval_loss": 3.114525556564331,
-      "eval_runtime": 308.5965,
-      "eval_samples_per_second": 35.759,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": 3.2367777824401855,
+      "eval_runtime": 323.3817,
+      "eval_samples_per_second": 34.124,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 4200
     },
     {
       "epoch": 4.11,
-      "learning_rate": 0.0006475877309434917,
-      "loss": 2.9197,
+      "learning_rate": 0.0066950064678338835,
+      "loss": 3.2188,
       "step": 4500
     },
     {
       "epoch": 4.11,
-      "eval_loss": 3.0324432849884033,
-      "eval_runtime": 308.9122,
-      "eval_samples_per_second": 35.722,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": 3.2376551628112793,
+      "eval_runtime": 323.9231,
+      "eval_samples_per_second": 34.067,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 4500
     },
     {
       "epoch": 4.38,
-      "eval_loss": 2.999401807785034,
-      "eval_runtime": 308.612,
-      "eval_samples_per_second": 35.757,
-      "eval_steps_per_second": 1.118,
-      "eval_wer": 1.9598550067965563,
+      "eval_loss": 3.4206786155700684,
+      "eval_runtime": 324.1643,
+      "eval_samples_per_second": 34.041,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 1.0,
       "step": 4800
     },
     {
       "epoch": 4.57,
-      "learning_rate": 0.0007196059394558908,
-      "loss": 2.9023,
+      "learning_rate": 0.007152364497567974,
+      "loss": 5.3067,
       "step": 5000
     },
     {
       "epoch": 4.66,
-      "eval_loss": 2.991722822189331,
-      "eval_runtime": 309.103,
-      "eval_samples_per_second": 35.7,
-      "eval_steps_per_second": 1.116,
-      "eval_wer": 1.8240144993203444,
+      "eval_loss": NaN,
+      "eval_runtime": 324.7836,
+      "eval_samples_per_second": 33.976,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 1.0,
       "step": 5100
     },
     {
       "epoch": 4.93,
-      "eval_loss": 2.9946165084838867,
-      "eval_runtime": 308.4148,
-      "eval_samples_per_second": 35.78,
-      "eval_steps_per_second": 1.119,
-      "eval_wer": 1.958948799275034,
+      "eval_loss": NaN,
+      "eval_runtime": 323.844,
+      "eval_samples_per_second": 34.075,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 1.0,
       "step": 5400
     },
     {
       "epoch": 5.02,
-      "learning_rate": 0.0006889164121067909,
-      "loss": 2.9007,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 5500
     },
     {
       "epoch": 5.21,
-      "eval_loss": 3.195502519607544,
-      "eval_runtime": 308.3861,
-      "eval_samples_per_second": 35.783,
-      "eval_steps_per_second": 1.119,
+      "eval_loss": NaN,
+      "eval_runtime": 323.0895,
+      "eval_samples_per_second": 34.155,
+      "eval_steps_per_second": 1.068,
       "eval_wer": 1.0,
       "step": 5700
     },
     {
       "epoch": 5.48,
-      "learning_rate": 0.0006574616321963427,
-      "loss": 3.1887,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 6000
     },
     {
       "epoch": 5.48,
-      "eval_loss": 3.1901698112487793,
-      "eval_runtime": 308.4713,
-      "eval_samples_per_second": 35.773,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": NaN,
+      "eval_runtime": 323.9944,
+      "eval_samples_per_second": 34.059,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 6000
     },
     {
       "epoch": 5.75,
-      "eval_loss": 3.167245864868164,
-      "eval_runtime": 308.5576,
-      "eval_samples_per_second": 35.763,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": NaN,
+      "eval_runtime": 324.3089,
+      "eval_samples_per_second": 34.026,
+      "eval_steps_per_second": 1.064,
       "eval_wer": 1.0,
       "step": 6300
     },
     {
       "epoch": 5.94,
-      "learning_rate": 0.0006259438166548115,
-      "loss": 3.135,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 6500
     },
     {
       "epoch": 6.03,
-      "eval_loss": 3.2076234817504883,
-      "eval_runtime": 308.784,
-      "eval_samples_per_second": 35.737,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 323.6949,
+      "eval_samples_per_second": 34.091,
+      "eval_steps_per_second": 1.066,
       "eval_wer": 1.0,
       "step": 6600
     },
     {
       "epoch": 6.3,
-      "eval_loss": 3.212040424346924,
-      "eval_runtime": 309.3257,
-      "eval_samples_per_second": 35.674,
-      "eval_steps_per_second": 1.115,
+      "eval_loss": NaN,
+      "eval_runtime": 323.4351,
+      "eval_samples_per_second": 34.118,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 6900
     },
     {
       "epoch": 6.39,
-      "learning_rate": 0.0005944260011132802,
-      "loss": 3.1482,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 7000
     },
     {
       "epoch": 6.58,
-      "eval_loss": 3.1832025051116943,
-      "eval_runtime": 308.2925,
-      "eval_samples_per_second": 35.794,
-      "eval_steps_per_second": 1.119,
+      "eval_loss": NaN,
+      "eval_runtime": 323.8786,
+      "eval_samples_per_second": 34.071,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 7200
     },
     {
       "epoch": 6.85,
-      "learning_rate": 0.0005629081855717488,
-      "loss": 3.1546,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 7500
     },
     {
       "epoch": 6.85,
-      "eval_loss": 3.1799111366271973,
-      "eval_runtime": 308.7059,
-      "eval_samples_per_second": 35.746,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": NaN,
+      "eval_runtime": 323.8341,
+      "eval_samples_per_second": 34.076,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 7500
     },
     {
       "epoch": 7.12,
-      "eval_loss": 3.2451581954956055,
-      "eval_runtime": 307.3881,
-      "eval_samples_per_second": 35.899,
-      "eval_steps_per_second": 1.122,
+      "eval_loss": NaN,
+      "eval_runtime": 323.3556,
+      "eval_samples_per_second": 34.127,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 7800
     },
     {
       "epoch": 7.31,
-      "learning_rate": 0.0005313903700302176,
-      "loss": 3.1567,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 8000
     },
     {
       "epoch": 7.4,
-      "eval_loss": 3.2318718433380127,
-      "eval_runtime": 308.2983,
-      "eval_samples_per_second": 35.793,
-      "eval_steps_per_second": 1.119,
+      "eval_loss": NaN,
+      "eval_runtime": 323.4333,
+      "eval_samples_per_second": 34.118,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 8100
     },
     {
       "epoch": 7.67,
-      "eval_loss": 3.222830057144165,
-      "eval_runtime": 308.3145,
-      "eval_samples_per_second": 35.791,
-      "eval_steps_per_second": 1.119,
+      "eval_loss": NaN,
+      "eval_runtime": 323.7546,
+      "eval_samples_per_second": 34.084,
+      "eval_steps_per_second": 1.066,
       "eval_wer": 1.0,
       "step": 8400
     },
     {
       "epoch": 7.76,
-      "learning_rate": 0.0004998725544886862,
-      "loss": 3.1719,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 8500
     },
     {
       "epoch": 7.95,
-      "eval_loss": 3.2054970264434814,
-      "eval_runtime": 308.0159,
-      "eval_samples_per_second": 35.826,
-      "eval_steps_per_second": 1.12,
+      "eval_loss": NaN,
+      "eval_runtime": 323.2121,
+      "eval_samples_per_second": 34.142,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 8700
     },
     {
       "epoch": 8.22,
-      "learning_rate": 0.00046835473894715497,
-      "loss": 3.168,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 9000
     },
     {
       "epoch": 8.22,
-      "eval_loss": 3.2552778720855713,
-      "eval_runtime": 303.2884,
-      "eval_samples_per_second": 36.385,
-      "eval_steps_per_second": 1.138,
+      "eval_loss": NaN,
+      "eval_runtime": 322.4572,
+      "eval_samples_per_second": 34.222,
+      "eval_steps_per_second": 1.07,
       "eval_wer": 1.0,
       "step": 9000
     },
     {
       "epoch": 8.49,
-      "eval_loss": 3.197523593902588,
-      "eval_runtime": 305.1147,
-      "eval_samples_per_second": 36.167,
-      "eval_steps_per_second": 1.131,
+      "eval_loss": NaN,
+      "eval_runtime": 318.5677,
+      "eval_samples_per_second": 34.639,
+      "eval_steps_per_second": 1.083,
       "eval_wer": 1.0,
       "step": 9300
     },
     {
       "epoch": 8.68,
-      "learning_rate": 0.0004368369234056237,
-      "loss": 3.1643,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 9500
     },
     {
       "epoch": 8.77,
-      "eval_loss": 3.2445874214172363,
-      "eval_runtime": 303.4296,
-      "eval_samples_per_second": 36.368,
-      "eval_steps_per_second": 1.137,
+      "eval_loss": NaN,
+      "eval_runtime": 319.654,
+      "eval_samples_per_second": 34.522,
+      "eval_steps_per_second": 1.079,
       "eval_wer": 1.0,
       "step": 9600
     },
     {
       "epoch": 9.04,
-      "eval_loss": 3.2781076431274414,
-      "eval_runtime": 305.7237,
-      "eval_samples_per_second": 36.095,
-      "eval_steps_per_second": 1.128,
+      "eval_loss": NaN,
+      "eval_runtime": 320.9193,
+      "eval_samples_per_second": 34.386,
+      "eval_steps_per_second": 1.075,
       "eval_wer": 1.0,
       "step": 9900
     },
     {
       "epoch": 9.13,
-      "learning_rate": 0.0004053191078640924,
-      "loss": 3.169,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 10000
     },
     {
       "epoch": 9.32,
-      "eval_loss": 3.2596964836120605,
-      "eval_runtime": 306.6385,
-      "eval_samples_per_second": 35.987,
-      "eval_steps_per_second": 1.125,
+      "eval_loss": NaN,
+      "eval_runtime": 322.8537,
+      "eval_samples_per_second": 34.18,
+      "eval_steps_per_second": 1.069,
       "eval_wer": 1.0,
       "step": 10200
     },
     {
       "epoch": 9.59,
-      "learning_rate": 0.00037380129232256106,
-      "loss": 3.1789,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 10500
     },
     {
       "epoch": 9.59,
-      "eval_loss": 3.2585501670837402,
-      "eval_runtime": 307.6539,
-      "eval_samples_per_second": 35.868,
-      "eval_steps_per_second": 1.121,
+      "eval_loss": NaN,
+      "eval_runtime": 321.9757,
+      "eval_samples_per_second": 34.273,
+      "eval_steps_per_second": 1.072,
       "eval_wer": 1.0,
       "step": 10500
     },
     {
       "epoch": 9.86,
-      "eval_loss": 3.2689764499664307,
-      "eval_runtime": 307.738,
-      "eval_samples_per_second": 35.858,
-      "eval_steps_per_second": 1.121,
+      "eval_loss": NaN,
+      "eval_runtime": 323.2439,
+      "eval_samples_per_second": 34.138,
+      "eval_steps_per_second": 1.067,
       "eval_wer": 1.0,
       "step": 10800
     },
     {
       "epoch": 10.05,
-      "learning_rate": 0.0003422834767810298,
-      "loss": 3.1701,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 11000
     },
     {
       "epoch": 10.14,
-      "eval_loss": 3.273723602294922,
-      "eval_runtime": 308.5346,
-      "eval_samples_per_second": 35.766,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": NaN,
+      "eval_runtime": 323.9537,
+      "eval_samples_per_second": 34.064,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 11100
     },
     {
       "epoch": 10.41,
-      "eval_loss": 3.273848533630371,
-      "eval_runtime": 308.8614,
-      "eval_samples_per_second": 35.728,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 323.8438,
+      "eval_samples_per_second": 34.075,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 11400
     },
     {
       "epoch": 10.5,
-      "learning_rate": 0.00031076566123949855,
-      "loss": 3.1698,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 11500
     },
     {
       "epoch": 10.68,
-      "eval_loss": 3.2595293521881104,
-      "eval_runtime": 308.7628,
-      "eval_samples_per_second": 35.739,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 323.9188,
+      "eval_samples_per_second": 34.067,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 11700
     },
     {
       "epoch": 10.96,
-      "learning_rate": 0.00027924784569796727,
-      "loss": 3.1595,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 12000
     },
     {
       "epoch": 10.96,
-      "eval_loss": 3.2467362880706787,
-      "eval_runtime": 308.3094,
-      "eval_samples_per_second": 35.792,
-      "eval_steps_per_second": 1.119,
+      "eval_loss": NaN,
+      "eval_runtime": 323.8826,
+      "eval_samples_per_second": 34.071,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 12000
     },
     {
       "epoch": 11.23,
-      "eval_loss": 3.252420663833618,
-      "eval_runtime": 309.0904,
-      "eval_samples_per_second": 35.702,
-      "eval_steps_per_second": 1.116,
+      "eval_loss": NaN,
+      "eval_runtime": 325.2969,
+      "eval_samples_per_second": 33.923,
+      "eval_steps_per_second": 1.061,
       "eval_wer": 1.0,
       "step": 12300
     },
     {
       "epoch": 11.42,
-      "learning_rate": 0.00024773003015643593,
-      "loss": 3.15,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 12500
     },
     {
       "epoch": 11.51,
-      "eval_loss": 3.2327377796173096,
-      "eval_runtime": 308.9397,
-      "eval_samples_per_second": 35.719,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 324.7333,
+      "eval_samples_per_second": 33.982,
+      "eval_steps_per_second": 1.062,
       "eval_wer": 1.0,
       "step": 12600
     },
     {
       "epoch": 11.78,
-      "eval_loss": 3.219557046890259,
-      "eval_runtime": 309.2594,
-      "eval_samples_per_second": 35.682,
-      "eval_steps_per_second": 1.116,
+      "eval_loss": NaN,
+      "eval_runtime": 324.3204,
+      "eval_samples_per_second": 34.025,
+      "eval_steps_per_second": 1.064,
       "eval_wer": 1.0,
       "step": 12900
     },
     {
       "epoch": 11.87,
-      "learning_rate": 0.00021621221461490465,
-      "loss": 3.1444,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 13000
     },
     {
       "epoch": 12.05,
-      "eval_loss": 3.1942968368530273,
-      "eval_runtime": 309.871,
-      "eval_samples_per_second": 35.612,
-      "eval_steps_per_second": 1.113,
+      "eval_loss": NaN,
+      "eval_runtime": 324.8823,
+      "eval_samples_per_second": 33.966,
+      "eval_steps_per_second": 1.062,
       "eval_wer": 1.0,
       "step": 13200
     },
     {
       "epoch": 12.33,
-      "learning_rate": 0.00018469439907337336,
-      "loss": 3.132,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 13500
     },
     {
       "epoch": 12.33,
-      "eval_loss": 3.191138744354248,
-      "eval_runtime": 309.3206,
-      "eval_samples_per_second": 35.675,
-      "eval_steps_per_second": 1.115,
+      "eval_loss": NaN,
+      "eval_runtime": 323.9396,
+      "eval_samples_per_second": 34.065,
+      "eval_steps_per_second": 1.065,
       "eval_wer": 1.0,
       "step": 13500
     },
     {
       "epoch": 12.6,
-      "eval_loss": 3.207465648651123,
-      "eval_runtime": 309.5517,
-      "eval_samples_per_second": 35.648,
-      "eval_steps_per_second": 1.115,
+      "eval_loss": NaN,
+      "eval_runtime": 326.2533,
+      "eval_samples_per_second": 33.823,
+      "eval_steps_per_second": 1.057,
       "eval_wer": 1.0,
       "step": 13800
     },
     {
       "epoch": 12.79,
-      "learning_rate": 0.00015323961916292511,
-      "loss": 3.1153,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 14000
     },
     {
       "epoch": 12.88,
-      "eval_loss": 3.1938300132751465,
-      "eval_runtime": 310.0376,
-      "eval_samples_per_second": 35.592,
-      "eval_steps_per_second": 1.113,
+      "eval_loss": NaN,
+      "eval_runtime": 323.7094,
+      "eval_samples_per_second": 34.089,
+      "eval_steps_per_second": 1.066,
       "eval_wer": 1.0,
       "step": 14100
     },
     {
       "epoch": 13.15,
-      "eval_loss": 3.1638731956481934,
-      "eval_runtime": 308.9592,
-      "eval_samples_per_second": 35.717,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 325.3499,
+      "eval_samples_per_second": 33.917,
+      "eval_steps_per_second": 1.06,
       "eval_wer": 1.0,
       "step": 14400
     },
     {
       "epoch": 13.24,
-      "learning_rate": 0.00012172180362139385,
-      "loss": 3.1039,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 14500
     },
     {
       "epoch": 13.42,
-      "eval_loss": 3.15146803855896,
-      "eval_runtime": 308.6922,
-      "eval_samples_per_second": 35.748,
-      "eval_steps_per_second": 1.118,
+      "eval_loss": NaN,
+      "eval_runtime": 325.7039,
+      "eval_samples_per_second": 33.88,
+      "eval_steps_per_second": 1.059,
       "eval_wer": 1.0,
       "step": 14700
     },
     {
       "epoch": 13.7,
-      "learning_rate": 9.020398807986256e-05,
-      "loss": 3.0839,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 15000
     },
     {
       "epoch": 13.7,
-      "eval_loss": 3.153453826904297,
-      "eval_runtime": 309.1197,
-      "eval_samples_per_second": 35.698,
-      "eval_steps_per_second": 1.116,
+      "eval_loss": NaN,
+      "eval_runtime": 323.516,
+      "eval_samples_per_second": 34.11,
+      "eval_steps_per_second": 1.066,
       "eval_wer": 1.0,
       "step": 15000
     },
     {
       "epoch": 13.97,
-      "eval_loss": 3.130723237991333,
-      "eval_runtime": 309.6167,
-      "eval_samples_per_second": 35.641,
-      "eval_steps_per_second": 1.114,
+      "eval_loss": NaN,
+      "eval_runtime": 471.9655,
+      "eval_samples_per_second": 23.381,
+      "eval_steps_per_second": 0.731,
       "eval_wer": 1.0,
       "step": 15300
     },
     {
       "epoch": 14.16,
-      "learning_rate": 5.8686172538331265e-05,
-      "loss": 3.0632,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 15500
     },
     {
       "epoch": 14.25,
-      "eval_loss": 3.1138317584991455,
-      "eval_runtime": 309.4562,
-      "eval_samples_per_second": 35.659,
-      "eval_steps_per_second": 1.115,
+      "eval_loss": NaN,
+      "eval_runtime": 324.2368,
+      "eval_samples_per_second": 34.034,
+      "eval_steps_per_second": 1.064,
       "eval_wer": 1.0,
       "step": 15600
     },
     {
       "epoch": 14.52,
-      "eval_loss": 3.128912925720215,
-      "eval_runtime": 309.4874,
-      "eval_samples_per_second": 35.656,
-      "eval_steps_per_second": 1.115,
+      "eval_loss": NaN,
+      "eval_runtime": 325.6971,
+      "eval_samples_per_second": 33.881,
+      "eval_steps_per_second": 1.059,
       "eval_wer": 1.0,
       "step": 15900
     },
     {
       "epoch": 14.61,
-      "learning_rate": 2.7168356996799972e-05,
-      "loss": 3.0518,
+      "learning_rate": 0.007152364497567974,
+      "loss": 0.0,
       "step": 16000
     },
     {
       "epoch": 14.79,
-      "eval_loss": 3.081491708755493,
-      "eval_runtime": 308.8218,
-      "eval_samples_per_second": 35.733,
-      "eval_steps_per_second": 1.117,
+      "eval_loss": NaN,
+      "eval_runtime": 325.4219,
+      "eval_samples_per_second": 33.91,
+      "eval_steps_per_second": 1.06,
       "eval_wer": 1.0,
       "step": 16200
     },
@@ -690,10 +690,10 @@
       "epoch": 15.0,
       "step": 16425,
       "total_flos": 6.442470243808035e+19,
-      "train_loss": 3.3253096312547563,
-      "train_runtime": 44962.3834,
-      "train_samples_per_second": 14.612,
-      "train_steps_per_second": 0.365
+      "train_loss": 1.129231213434646,
+      "train_runtime": 45647.7738,
+      "train_samples_per_second": 14.392,
+      "train_steps_per_second": 0.36
     }
   ],
   "logging_steps": 500,