End of training

README.md

@@ -1,8 +1,12 @@
 ---
 library_name: transformers
+language:
+- ja
 license: apache-2.0
 base_model: rinna/japanese-hubert-base
 tags:
+- automatic-speech-recognition
+- mozilla-foundation/common_voice_13_0
 - generated_from_trainer
 datasets:
 - common_voice_13_0
@@ -15,11 +19,11 @@ model-index:
       name: Automatic Speech Recognition
       type: automatic-speech-recognition
     dataset:
-      name: common_voice_13_0
+      name: MOZILLA-FOUNDATION/COMMON_VOICE_13_0 - JA
       type: common_voice_13_0
       config: ja
       split: test
-      args: ja
+      args: 'Config: ja, Training split: train+validation, Eval split: test'
     metrics:
     - name: Wer
       type: wer
@@ -31,11 +35,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Hubert-common_voice-ja-demo-phonemes-cosine-3e-5
 
-This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the common_voice_13_0 dataset.
+This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the MOZILLA-FOUNDATION/COMMON_VOICE_13_0 - JA dataset.
 It achieves the following results on the evaluation set:
 - Loss: inf
 - Wer: 1.0
-- Cer: 0.2334
+- Cer: 0.2359
 
 ## Model description
 

all_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 20.0,
+    "eval_cer": 0.23587889938512407,
+    "eval_loss": Infinity,
+    "eval_runtime": 211.3713,
+    "eval_samples": 4961,
+    "eval_samples_per_second": 23.471,
+    "eval_steps_per_second": 2.938,
+    "eval_wer": 1.0,
+    "total_flos": 1.051128494332674e+19,
+    "train_loss": 3.393639066751967,
+    "train_runtime": 40261.649,
+    "train_samples": 12032,
+    "train_samples_per_second": 5.977,
+    "train_steps_per_second": 0.187
+}

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 20.0,
+    "eval_cer": 0.23587889938512407,
+    "eval_loss": Infinity,
+    "eval_runtime": 211.3713,
+    "eval_samples": 4961,
+    "eval_samples_per_second": 23.471,
+    "eval_steps_per_second": 2.938,
+    "eval_wer": 1.0
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 20.0,
+    "total_flos": 1.051128494332674e+19,
+    "train_loss": 3.393639066751967,
+    "train_runtime": 40261.649,
+    "train_samples": 12032,
+    "train_samples_per_second": 5.977,
+    "train_steps_per_second": 0.187
+}

trainer_state.json

@@ -0,0 +1,897 @@
+{
+  "best_metric": 0.6672275066375732,
+  "best_model_checkpoint": "./Hubert-common_voice-ja-demo-phonemes-cosine-3e-5/checkpoint-6100",
+  "epoch": 20.0,
+  "eval_steps": 100,
+  "global_step": 7520,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.26595744680851063,
+      "eval_cer": 4.906748881550685,
+      "eval_loss": Infinity,
+      "eval_runtime": 231.9839,
+      "eval_samples_per_second": 21.385,
+      "eval_steps_per_second": 2.677,
+      "eval_wer": 1.8203991130820398,
+      "step": 100
+    },
+    {
+      "epoch": 0.5319148936170213,
+      "eval_cer": 4.632337179381479,
+      "eval_loss": Infinity,
+      "eval_runtime": 212.1645,
+      "eval_samples_per_second": 23.383,
+      "eval_steps_per_second": 2.927,
+      "eval_wer": 1.5926224551501713,
+      "step": 200
+    },
+    {
+      "epoch": 0.7978723404255319,
+      "eval_cer": 1.9636802298082472,
+      "eval_loss": Infinity,
+      "eval_runtime": 212.1688,
+      "eval_samples_per_second": 23.382,
+      "eval_steps_per_second": 2.927,
+      "eval_wer": 1.1769804474904253,
+      "step": 300
+    },
+    {
+      "epoch": 1.0638297872340425,
+      "eval_cer": 0.9816665437204115,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.9606,
+      "eval_samples_per_second": 23.741,
+      "eval_steps_per_second": 2.972,
+      "eval_wer": 1.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.3297872340425532,
+      "grad_norm": 50.47308349609375,
+      "learning_rate": 1.1904e-06,
+      "loss": 14.493,
+      "step": 500
+    },
+    {
+      "epoch": 1.3297872340425532,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.2095,
+      "eval_samples_per_second": 23.713,
+      "eval_steps_per_second": 2.968,
+      "eval_wer": 1.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.5957446808510638,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.2442,
+      "eval_samples_per_second": 23.823,
+      "eval_steps_per_second": 2.982,
+      "eval_wer": 1.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.8617021276595744,
+      "eval_cer": 0.9816622044408188,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.3548,
+      "eval_samples_per_second": 23.81,
+      "eval_steps_per_second": 2.98,
+      "eval_wer": 1.0,
+      "step": 700
+    },
+    {
+      "epoch": 2.127659574468085,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 207.0664,
+      "eval_samples_per_second": 23.958,
+      "eval_steps_per_second": 2.999,
+      "eval_wer": 1.0,
+      "step": 800
+    },
+    {
+      "epoch": 2.393617021276596,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": Infinity,
+      "eval_runtime": 207.7036,
+      "eval_samples_per_second": 23.885,
+      "eval_steps_per_second": 2.99,
+      "eval_wer": 1.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.6595744680851063,
+      "grad_norm": 46.10651397705078,
+      "learning_rate": 2.3904e-06,
+      "loss": 6.5744,
+      "step": 1000
+    },
+    {
+      "epoch": 2.6595744680851063,
+      "eval_cer": 0.9816839008387828,
+      "eval_loss": 6.808023452758789,
+      "eval_runtime": 220.5662,
+      "eval_samples_per_second": 22.492,
+      "eval_steps_per_second": 2.815,
+      "eval_wer": 1.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.925531914893617,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": 6.59719705581665,
+      "eval_runtime": 208.0758,
+      "eval_samples_per_second": 23.842,
+      "eval_steps_per_second": 2.984,
+      "eval_wer": 1.0,
+      "step": 1100
+    },
+    {
+      "epoch": 3.1914893617021276,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": Infinity,
+      "eval_runtime": 207.6614,
+      "eval_samples_per_second": 23.89,
+      "eval_steps_per_second": 2.99,
+      "eval_wer": 1.0,
+      "step": 1200
+    },
+    {
+      "epoch": 3.4574468085106385,
+      "eval_cer": 0.9816882401183755,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.3327,
+      "eval_samples_per_second": 23.699,
+      "eval_steps_per_second": 2.967,
+      "eval_wer": 1.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.723404255319149,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.2723,
+      "eval_samples_per_second": 23.82,
+      "eval_steps_per_second": 2.982,
+      "eval_wer": 1.0,
+      "step": 1400
+    },
+    {
+      "epoch": 3.9893617021276597,
+      "grad_norm": 31.730201721191406,
+      "learning_rate": 3.5904e-06,
+      "loss": 5.5193,
+      "step": 1500
+    },
+    {
+      "epoch": 3.9893617021276597,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.1441,
+      "eval_samples_per_second": 23.834,
+      "eval_steps_per_second": 2.984,
+      "eval_wer": 1.0,
+      "step": 1500
+    },
+    {
+      "epoch": 4.25531914893617,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 211.1492,
+      "eval_samples_per_second": 23.495,
+      "eval_steps_per_second": 2.941,
+      "eval_wer": 1.0,
+      "step": 1600
+    },
+    {
+      "epoch": 4.5212765957446805,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.0326,
+      "eval_samples_per_second": 23.847,
+      "eval_steps_per_second": 2.985,
+      "eval_wer": 1.0,
+      "step": 1700
+    },
+    {
+      "epoch": 4.787234042553192,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 212.7439,
+      "eval_samples_per_second": 23.319,
+      "eval_steps_per_second": 2.919,
+      "eval_wer": 1.0,
+      "step": 1800
+    },
+    {
+      "epoch": 5.053191489361702,
+      "eval_cer": 0.9816622044408188,
+      "eval_loss": Infinity,
+      "eval_runtime": 211.9236,
+      "eval_samples_per_second": 23.409,
+      "eval_steps_per_second": 2.93,
+      "eval_wer": 1.0,
+      "step": 1900
+    },
+    {
+      "epoch": 5.319148936170213,
+      "grad_norm": 14.480951309204102,
+      "learning_rate": 4.7904e-06,
+      "loss": 4.5578,
+      "step": 2000
+    },
+    {
+      "epoch": 5.319148936170213,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 225.0981,
+      "eval_samples_per_second": 22.039,
+      "eval_steps_per_second": 2.759,
+      "eval_wer": 1.0,
+      "step": 2000
+    },
+    {
+      "epoch": 5.585106382978723,
+      "eval_cer": 0.9816622044408188,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.2028,
+      "eval_samples_per_second": 23.828,
+      "eval_steps_per_second": 2.983,
+      "eval_wer": 1.0,
+      "step": 2100
+    },
+    {
+      "epoch": 5.851063829787234,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.7044,
+      "eval_samples_per_second": 23.77,
+      "eval_steps_per_second": 2.976,
+      "eval_wer": 1.0,
+      "step": 2200
+    },
+    {
+      "epoch": 6.117021276595745,
+      "eval_cer": 0.9816665437204115,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.3105,
+      "eval_samples_per_second": 23.815,
+      "eval_steps_per_second": 2.981,
+      "eval_wer": 1.0,
+      "step": 2300
+    },
+    {
+      "epoch": 6.382978723404255,
+      "eval_cer": 0.9816839008387828,
+      "eval_loss": Infinity,
+      "eval_runtime": 208.6388,
+      "eval_samples_per_second": 23.778,
+      "eval_steps_per_second": 2.976,
+      "eval_wer": 1.0,
+      "step": 2400
+    },
+    {
+      "epoch": 6.648936170212766,
+      "grad_norm": 10.711950302124023,
+      "learning_rate": 5.9904e-06,
+      "loss": 3.6943,
+      "step": 2500
+    },
+    {
+      "epoch": 6.648936170212766,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.8055,
+      "eval_samples_per_second": 23.646,
+      "eval_steps_per_second": 2.96,
+      "eval_wer": 1.0,
+      "step": 2500
+    },
+    {
+      "epoch": 6.914893617021277,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.752,
+      "eval_samples_per_second": 23.652,
+      "eval_steps_per_second": 2.961,
+      "eval_wer": 1.0,
+      "step": 2600
+    },
+    {
+      "epoch": 7.180851063829787,
+      "eval_cer": 0.9816839008387828,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.5647,
+      "eval_samples_per_second": 23.673,
+      "eval_steps_per_second": 2.963,
+      "eval_wer": 1.0,
+      "step": 2700
+    },
+    {
+      "epoch": 7.446808510638298,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.0972,
+      "eval_samples_per_second": 23.726,
+      "eval_steps_per_second": 2.97,
+      "eval_wer": 1.0,
+      "step": 2800
+    },
+    {
+      "epoch": 7.712765957446808,
+      "eval_cer": 0.9816665437204115,
+      "eval_loss": 3.1572225093841553,
+      "eval_runtime": 214.7616,
+      "eval_samples_per_second": 23.1,
+      "eval_steps_per_second": 2.892,
+      "eval_wer": 1.0,
+      "step": 2900
+    },
+    {
+      "epoch": 7.9787234042553195,
+      "grad_norm": 3.429447650909424,
+      "learning_rate": 7.190400000000001e-06,
+      "loss": 3.1932,
+      "step": 3000
+    },
+    {
+      "epoch": 7.9787234042553195,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.8411,
+      "eval_samples_per_second": 23.642,
+      "eval_steps_per_second": 2.959,
+      "eval_wer": 1.0,
+      "step": 3000
+    },
+    {
+      "epoch": 8.24468085106383,
+      "eval_cer": 0.9816795615591899,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.153,
+      "eval_samples_per_second": 23.607,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 3100
+    },
+    {
+      "epoch": 8.51063829787234,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.9276,
+      "eval_samples_per_second": 23.632,
+      "eval_steps_per_second": 2.958,
+      "eval_wer": 1.0,
+      "step": 3200
+    },
+    {
+      "epoch": 8.77659574468085,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.788,
+      "eval_samples_per_second": 23.648,
+      "eval_steps_per_second": 2.96,
+      "eval_wer": 1.0,
+      "step": 3300
+    },
+    {
+      "epoch": 9.042553191489361,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.3829,
+      "eval_samples_per_second": 23.693,
+      "eval_steps_per_second": 2.966,
+      "eval_wer": 1.0,
+      "step": 3400
+    },
+    {
+      "epoch": 9.308510638297872,
+      "grad_norm": 0.5232295393943787,
+      "learning_rate": 8.3904e-06,
+      "loss": 3.0309,
+      "step": 3500
+    },
+    {
+      "epoch": 9.308510638297872,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.6347,
+      "eval_samples_per_second": 23.665,
+      "eval_steps_per_second": 2.962,
+      "eval_wer": 1.0,
+      "step": 3500
+    },
+    {
+      "epoch": 9.574468085106384,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.0039,
+      "eval_samples_per_second": 23.736,
+      "eval_steps_per_second": 2.971,
+      "eval_wer": 1.0,
+      "step": 3600
+    },
+    {
+      "epoch": 9.840425531914894,
+      "eval_cer": 0.9816622044408188,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.5465,
+      "eval_samples_per_second": 23.675,
+      "eval_steps_per_second": 2.964,
+      "eval_wer": 1.0,
+      "step": 3700
+    },
+    {
+      "epoch": 10.106382978723405,
+      "eval_cer": 0.9816752222795971,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.1604,
+      "eval_samples_per_second": 23.719,
+      "eval_steps_per_second": 2.969,
+      "eval_wer": 1.0,
+      "step": 3800
+    },
+    {
+      "epoch": 10.372340425531915,
+      "eval_cer": 0.9816665437204115,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.698,
+      "eval_samples_per_second": 23.658,
+      "eval_steps_per_second": 2.961,
+      "eval_wer": 1.0,
+      "step": 3900
+    },
+    {
+      "epoch": 10.638297872340425,
+      "grad_norm": 1.3846327066421509,
+      "learning_rate": 9.5904e-06,
+      "loss": 2.9704,
+      "step": 4000
+    },
+    {
+      "epoch": 10.638297872340425,
+      "eval_cer": 0.9816839008387828,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.1316,
+      "eval_samples_per_second": 23.609,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 4000
+    },
+    {
+      "epoch": 10.904255319148936,
+      "eval_cer": 0.9816708830000044,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.6004,
+      "eval_samples_per_second": 23.669,
+      "eval_steps_per_second": 2.963,
+      "eval_wer": 1.0,
+      "step": 4100
+    },
+    {
+      "epoch": 11.170212765957446,
+      "eval_cer": 0.9048569556482233,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.9962,
+      "eval_samples_per_second": 23.624,
+      "eval_steps_per_second": 2.957,
+      "eval_wer": 1.0,
+      "step": 4200
+    },
+    {
+      "epoch": 11.436170212765958,
+      "eval_cer": 0.7253886909695252,
+      "eval_loss": Infinity,
+      "eval_runtime": 209.5146,
+      "eval_samples_per_second": 23.679,
+      "eval_steps_per_second": 2.964,
+      "eval_wer": 1.0,
+      "step": 4300
+    },
+    {
+      "epoch": 11.702127659574469,
+      "eval_cer": 0.4365402055950671,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.2066,
+      "eval_samples_per_second": 23.601,
+      "eval_steps_per_second": 2.954,
+      "eval_wer": 1.0,
+      "step": 4400
+    },
+    {
+      "epoch": 11.96808510638298,
+      "grad_norm": 3.278918981552124,
+      "learning_rate": 1.0790400000000001e-05,
+      "loss": 2.2767,
+      "step": 4500
+    },
+    {
+      "epoch": 11.96808510638298,
+      "eval_cer": 0.3732301163360859,
+      "eval_loss": 1.5674688816070557,
+      "eval_runtime": 211.1485,
+      "eval_samples_per_second": 23.495,
+      "eval_steps_per_second": 2.941,
+      "eval_wer": 1.0,
+      "step": 4500
+    },
+    {
+      "epoch": 12.23404255319149,
+      "eval_cer": 0.3454630662217459,
+      "eval_loss": Infinity,
+      "eval_runtime": 211.2681,
+      "eval_samples_per_second": 23.482,
+      "eval_steps_per_second": 2.939,
+      "eval_wer": 1.0,
+      "step": 4600
+    },
+    {
+      "epoch": 12.5,
+      "eval_cer": 0.3277154126871857,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.9727,
+      "eval_samples_per_second": 23.515,
+      "eval_steps_per_second": 2.944,
+      "eval_wer": 1.0,
+      "step": 4700
+    },
+    {
+      "epoch": 12.76595744680851,
+      "eval_cer": 0.3052553015148425,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.721,
+      "eval_samples_per_second": 23.543,
+      "eval_steps_per_second": 2.947,
+      "eval_wer": 1.0,
+      "step": 4800
+    },
+    {
+      "epoch": 13.03191489361702,
+      "eval_cer": 0.2935305680550915,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.659,
+      "eval_samples_per_second": 23.55,
+      "eval_steps_per_second": 2.948,
+      "eval_wer": 1.0,
+      "step": 4900
+    },
+    {
+      "epoch": 13.297872340425531,
+      "grad_norm": 2.6736652851104736,
+      "learning_rate": 1.19904e-05,
+      "loss": 1.2873,
+      "step": 5000
+    },
+    {
+      "epoch": 13.297872340425531,
+      "eval_cer": 0.2783691251578413,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.6907,
+      "eval_samples_per_second": 23.546,
+      "eval_steps_per_second": 2.947,
+      "eval_wer": 1.0,
+      "step": 5000
+    },
+    {
+      "epoch": 13.563829787234042,
+      "eval_cer": 0.2684104784923607,
+      "eval_loss": Infinity,
+      "eval_runtime": 211.3579,
+      "eval_samples_per_second": 23.472,
+      "eval_steps_per_second": 2.938,
+      "eval_wer": 1.0,
+      "step": 5100
+    },
+    {
+      "epoch": 13.829787234042554,
+      "eval_cer": 0.26778996151059004,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.8829,
+      "eval_samples_per_second": 23.525,
+      "eval_steps_per_second": 2.945,
+      "eval_wer": 1.0,
+      "step": 5200
+    },
+    {
+      "epoch": 14.095744680851064,
+      "eval_cer": 0.26162818448881114,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.4536,
+      "eval_samples_per_second": 23.573,
+      "eval_steps_per_second": 2.951,
+      "eval_wer": 1.0,
+      "step": 5300
+    },
+    {
+      "epoch": 14.361702127659575,
+      "eval_cer": 0.2608384356029212,
+      "eval_loss": 0.8213518261909485,
+      "eval_runtime": 210.8924,
+      "eval_samples_per_second": 23.524,
+      "eval_steps_per_second": 2.945,
+      "eval_wer": 1.0,
+      "step": 5400
+    },
+    {
+      "epoch": 14.627659574468085,
+      "grad_norm": 4.707749843597412,
+      "learning_rate": 1.31904e-05,
+      "loss": 0.9318,
+      "step": 5500
+    },
+    {
+      "epoch": 14.627659574468085,
+      "eval_cer": 0.2563863347407063,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.577,
+      "eval_samples_per_second": 23.559,
+      "eval_steps_per_second": 2.949,
+      "eval_wer": 1.0,
+      "step": 5500
+    },
+    {
+      "epoch": 14.893617021276595,
+      "eval_cer": 0.25441630180557423,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.7551,
+      "eval_samples_per_second": 23.539,
+      "eval_steps_per_second": 2.947,
+      "eval_wer": 1.0,
+      "step": 5600
+    },
+    {
+      "epoch": 15.159574468085106,
+      "eval_cer": 0.25251135806433417,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.4666,
+      "eval_samples_per_second": 23.571,
+      "eval_steps_per_second": 2.951,
+      "eval_wer": 1.0,
+      "step": 5700
+    },
+    {
+      "epoch": 15.425531914893616,
+      "eval_cer": 0.25104034228237426,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.6399,
+      "eval_samples_per_second": 23.552,
+      "eval_steps_per_second": 2.948,
+      "eval_wer": 1.0,
+      "step": 5800
+    },
+    {
+      "epoch": 15.691489361702128,
+      "eval_cer": 0.25268058996845344,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.4046,
+      "eval_samples_per_second": 23.578,
+      "eval_steps_per_second": 2.951,
+      "eval_wer": 1.0,
+      "step": 5900
+    },
+    {
+      "epoch": 15.957446808510639,
+      "grad_norm": 3.4887096881866455,
+      "learning_rate": 1.43904e-05,
+      "loss": 0.754,
+      "step": 6000
+    },
+    {
+      "epoch": 15.957446808510639,
+      "eval_cer": 0.24993382598620978,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.2447,
+      "eval_samples_per_second": 23.596,
+      "eval_steps_per_second": 2.954,
+      "eval_wer": 1.0,
+      "step": 6000
+    },
+    {
+      "epoch": 16.22340425531915,
+      "eval_cer": 0.24853657795732753,
+      "eval_loss": 0.6672275066375732,
+      "eval_runtime": 210.4958,
+      "eval_samples_per_second": 23.568,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 1.0,
+      "step": 6100
+    },
+    {
+      "epoch": 16.48936170212766,
+      "eval_cer": 0.24641033095685455,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.6343,
+      "eval_samples_per_second": 23.553,
+      "eval_steps_per_second": 2.948,
+      "eval_wer": 1.0,
+      "step": 6200
+    },
+    {
+      "epoch": 16.75531914893617,
+      "eval_cer": 0.24667936629160828,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.3464,
+      "eval_samples_per_second": 23.585,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 1.0,
+      "step": 6300
+    },
+    {
+      "epoch": 17.02127659574468,
+      "eval_cer": 0.24105999921892968,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.367,
+      "eval_samples_per_second": 23.583,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 1.0,
+      "step": 6400
+    },
+    {
+      "epoch": 17.28723404255319,
+      "grad_norm": 3.309645652770996,
+      "learning_rate": 1.5590400000000002e-05,
+      "loss": 0.6421,
+      "step": 6500
+    },
+    {
+      "epoch": 17.28723404255319,
+      "eval_cer": 0.2411424455311929,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.1359,
+      "eval_samples_per_second": 23.609,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 6500
+    },
+    {
+      "epoch": 17.5531914893617,
+      "eval_cer": 0.2417846589109276,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.0208,
+      "eval_samples_per_second": 23.621,
+      "eval_steps_per_second": 2.957,
+      "eval_wer": 1.0,
+      "step": 6600
+    },
+    {
+      "epoch": 17.819148936170212,
+      "eval_cer": 0.23863868120614615,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.3543,
+      "eval_samples_per_second": 23.584,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 1.0,
+      "step": 6700
+    },
+    {
+      "epoch": 18.085106382978722,
+      "eval_cer": 0.23867773472248138,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.344,
+      "eval_samples_per_second": 23.585,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 0.999596855472687,
+      "step": 6800
+    },
+    {
+      "epoch": 18.351063829787233,
+      "eval_cer": 0.23813532477338112,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.5017,
+      "eval_samples_per_second": 23.568,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 1.0,
+      "step": 6900
+    },
+    {
+      "epoch": 18.617021276595743,
+      "grad_norm": 3.4141018390655518,
+      "learning_rate": 1.6790399999999998e-05,
+      "loss": 0.568,
+      "step": 7000
+    },
+    {
+      "epoch": 18.617021276595743,
+      "eval_cer": 0.23912468052053998,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.039,
+      "eval_samples_per_second": 23.619,
+      "eval_steps_per_second": 2.957,
+      "eval_wer": 1.0,
+      "step": 7000
+    },
+    {
+      "epoch": 18.882978723404257,
+      "eval_cer": 0.236976737122103,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.1426,
+      "eval_samples_per_second": 23.608,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 7100
+    },
+    {
+      "epoch": 19.148936170212767,
+      "eval_cer": 0.2344165621623498,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.1255,
+      "eval_samples_per_second": 23.61,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 7200
+    },
+    {
+      "epoch": 19.414893617021278,
+      "eval_cer": 0.2363562201403323,
+      "eval_loss": Infinity,
+      "eval_runtime": 211.0028,
+      "eval_samples_per_second": 23.512,
+      "eval_steps_per_second": 2.943,
+      "eval_wer": 1.0,
+      "step": 7300
+    },
+    {
+      "epoch": 19.680851063829788,
+      "eval_cer": 0.23471163317466034,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.5331,
+      "eval_samples_per_second": 23.564,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 1.0,
+      "step": 7400
+    },
+    {
+      "epoch": 19.9468085106383,
+      "grad_norm": 4.509326934814453,
+      "learning_rate": 1.79904e-05,
+      "loss": 0.5259,
+      "step": 7500
+    },
+    {
+      "epoch": 19.9468085106383,
+      "eval_cer": 0.23339249217844854,
+      "eval_loss": Infinity,
+      "eval_runtime": 210.6718,
+      "eval_samples_per_second": 23.548,
+      "eval_steps_per_second": 2.948,
+      "eval_wer": 1.0,
+      "step": 7500
+    },
+    {
+      "epoch": 20.0,
+      "step": 7520,
+      "total_flos": 1.051128494332674e+19,
+      "train_loss": 3.393639066751967,
+      "train_runtime": 40261.649,
+      "train_samples_per_second": 5.977,
+      "train_steps_per_second": 0.187
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 7520,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 400,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.051128494332674e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}