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
 metrics:
 - wer
@@ -16,11 +20,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Hubert-common_voice-phoneme-onlyJSUT
 
-This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the None 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: 0.1617
-- Wer: 0.998
-- Cer: 0.1060
+- Loss: 0.1563
+- Wer: 1.0
+- Cer: 0.1052
 
 ## Model description
 

all_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 20.0,
+    "eval_cer": 0.10522982265653276,
+    "eval_loss": 0.15634456276893616,
+    "eval_runtime": 20.6964,
+    "eval_samples": 500,
+    "eval_samples_per_second": 24.159,
+    "eval_steps_per_second": 3.044,
+    "eval_wer": 1.0,
+    "total_flos": 4.085369652431226e+18,
+    "train_loss": 2.2413718947282075,
+    "train_runtime": 5214.7646,
+    "train_samples": 4500,
+    "train_samples_per_second": 17.259,
+    "train_steps_per_second": 0.541
+}

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 20.0,
+    "eval_cer": 0.10522982265653276,
+    "eval_loss": 0.15634456276893616,
+    "eval_runtime": 20.6964,
+    "eval_samples": 500,
+    "eval_samples_per_second": 24.159,
+    "eval_steps_per_second": 3.044,
+    "eval_wer": 1.0
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 20.0,
+    "total_flos": 4.085369652431226e+18,
+    "train_loss": 2.2413718947282075,
+    "train_runtime": 5214.7646,
+    "train_samples": 4500,
+    "train_samples_per_second": 17.259,
+    "train_steps_per_second": 0.541
+}

trainer_state.json

@@ -0,0 +1,357 @@
+{
+  "best_metric": 0.16017669439315796,
+  "best_model_checkpoint": "./Hubert-common_voice-phoneme-onlyJSUT/checkpoint-2600",
+  "epoch": 20.0,
+  "eval_steps": 100,
+  "global_step": 2820,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.7092198581560284,
+      "eval_cer": 0.9860960308843045,
+      "eval_loss": 11.361393928527832,
+      "eval_runtime": 21.3659,
+      "eval_samples_per_second": 23.402,
+      "eval_steps_per_second": 2.949,
+      "eval_wer": 1.054,
+      "step": 100
+    },
+    {
+      "epoch": 1.4184397163120568,
+      "eval_cer": 0.9851007359150682,
+      "eval_loss": 5.935764789581299,
+      "eval_runtime": 21.3202,
+      "eval_samples_per_second": 23.452,
+      "eval_steps_per_second": 2.955,
+      "eval_wer": 1.0,
+      "step": 200
+    },
+    {
+      "epoch": 2.127659574468085,
+      "eval_cer": 0.9850705754614549,
+      "eval_loss": 5.310074329376221,
+      "eval_runtime": 20.7223,
+      "eval_samples_per_second": 24.129,
+      "eval_steps_per_second": 3.04,
+      "eval_wer": 1.0,
+      "step": 300
+    },
+    {
+      "epoch": 2.8368794326241136,
+      "eval_cer": 0.9851007359150682,
+      "eval_loss": 4.895324230194092,
+      "eval_runtime": 20.9802,
+      "eval_samples_per_second": 23.832,
+      "eval_steps_per_second": 3.003,
+      "eval_wer": 1.0,
+      "step": 400
+    },
+    {
+      "epoch": 3.546099290780142,
+      "grad_norm": 18.17830467224121,
+      "learning_rate": 1.1903999999999998e-05,
+      "loss": 6.9061,
+      "step": 500
+    },
+    {
+      "epoch": 3.546099290780142,
+      "eval_cer": 0.9851007359150682,
+      "eval_loss": 4.402055740356445,
+      "eval_runtime": 21.0039,
+      "eval_samples_per_second": 23.805,
+      "eval_steps_per_second": 2.999,
+      "eval_wer": 1.0,
+      "step": 500
+    },
+    {
+      "epoch": 4.25531914893617,
+      "eval_cer": 0.9851007359150682,
+      "eval_loss": 3.9323222637176514,
+      "eval_runtime": 21.0838,
+      "eval_samples_per_second": 23.715,
+      "eval_steps_per_second": 2.988,
+      "eval_wer": 1.0,
+      "step": 600
+    },
+    {
+      "epoch": 4.964539007092198,
+      "eval_cer": 0.9851007359150682,
+      "eval_loss": 3.4932363033294678,
+      "eval_runtime": 20.4799,
+      "eval_samples_per_second": 24.414,
+      "eval_steps_per_second": 3.076,
+      "eval_wer": 1.0,
+      "step": 700
+    },
+    {
+      "epoch": 5.673758865248227,
+      "eval_cer": 0.9850404150078417,
+      "eval_loss": 3.209160566329956,
+      "eval_runtime": 21.0193,
+      "eval_samples_per_second": 23.788,
+      "eval_steps_per_second": 2.997,
+      "eval_wer": 1.0,
+      "step": 800
+    },
+    {
+      "epoch": 6.382978723404255,
+      "eval_cer": 0.9850705754614549,
+      "eval_loss": 3.0484230518341064,
+      "eval_runtime": 20.7411,
+      "eval_samples_per_second": 24.107,
+      "eval_steps_per_second": 3.037,
+      "eval_wer": 1.0,
+      "step": 900
+    },
+    {
+      "epoch": 7.092198581560283,
+      "grad_norm": 0.6846523284912109,
+      "learning_rate": 2.3903999999999997e-05,
+      "loss": 3.4303,
+      "step": 1000
+    },
+    {
+      "epoch": 7.092198581560283,
+      "eval_cer": 0.9850404150078417,
+      "eval_loss": 2.9960806369781494,
+      "eval_runtime": 20.4215,
+      "eval_samples_per_second": 24.484,
+      "eval_steps_per_second": 3.085,
+      "eval_wer": 1.0,
+      "step": 1000
+    },
+    {
+      "epoch": 7.801418439716312,
+      "eval_cer": 0.9850102545542285,
+      "eval_loss": 2.799971580505371,
+      "eval_runtime": 20.9162,
+      "eval_samples_per_second": 23.905,
+      "eval_steps_per_second": 3.012,
+      "eval_wer": 1.0,
+      "step": 1100
+    },
+    {
+      "epoch": 8.51063829787234,
+      "eval_cer": 0.5949149475208108,
+      "eval_loss": 1.9060593843460083,
+      "eval_runtime": 21.115,
+      "eval_samples_per_second": 23.68,
+      "eval_steps_per_second": 2.984,
+      "eval_wer": 1.0,
+      "step": 1200
+    },
+    {
+      "epoch": 9.21985815602837,
+      "eval_cer": 0.1546929665822174,
+      "eval_loss": 0.8767441511154175,
+      "eval_runtime": 20.8067,
+      "eval_samples_per_second": 24.031,
+      "eval_steps_per_second": 3.028,
+      "eval_wer": 1.0,
+      "step": 1300
+    },
+    {
+      "epoch": 9.929078014184396,
+      "eval_cer": 0.12679454698998674,
+      "eval_loss": 0.5386155843734741,
+      "eval_runtime": 21.0364,
+      "eval_samples_per_second": 23.768,
+      "eval_steps_per_second": 2.995,
+      "eval_wer": 1.0,
+      "step": 1400
+    },
+    {
+      "epoch": 10.638297872340425,
+      "grad_norm": 2.450634241104126,
+      "learning_rate": 3.5903999999999994e-05,
+      "loss": 1.6163,
+      "step": 1500
+    },
+    {
+      "epoch": 10.638297872340425,
+      "eval_cer": 0.11901314995777537,
+      "eval_loss": 0.38196861743927,
+      "eval_runtime": 21.1077,
+      "eval_samples_per_second": 23.688,
+      "eval_steps_per_second": 2.985,
+      "eval_wer": 1.0,
+      "step": 1500
+    },
+    {
+      "epoch": 11.347517730496454,
+      "eval_cer": 0.11382555193630112,
+      "eval_loss": 0.298287034034729,
+      "eval_runtime": 21.1084,
+      "eval_samples_per_second": 23.687,
+      "eval_steps_per_second": 2.985,
+      "eval_wer": 1.0,
+      "step": 1600
+    },
+    {
+      "epoch": 12.056737588652481,
+      "eval_cer": 0.11165399927614911,
+      "eval_loss": 0.25243452191352844,
+      "eval_runtime": 20.9043,
+      "eval_samples_per_second": 23.919,
+      "eval_steps_per_second": 3.014,
+      "eval_wer": 1.0,
+      "step": 1700
+    },
+    {
+      "epoch": 12.76595744680851,
+      "eval_cer": 0.110417420678007,
+      "eval_loss": 0.22604165971279144,
+      "eval_runtime": 20.6318,
+      "eval_samples_per_second": 24.234,
+      "eval_steps_per_second": 3.054,
+      "eval_wer": 1.0,
+      "step": 1800
+    },
+    {
+      "epoch": 13.47517730496454,
+      "eval_cer": 0.11099046929665822,
+      "eval_loss": 0.20960931479930878,
+      "eval_runtime": 20.9655,
+      "eval_samples_per_second": 23.849,
+      "eval_steps_per_second": 3.005,
+      "eval_wer": 1.0,
+      "step": 1900
+    },
+    {
+      "epoch": 14.184397163120567,
+      "grad_norm": 1.458462119102478,
+      "learning_rate": 4.7903999999999994e-05,
+      "loss": 0.332,
+      "step": 2000
+    },
+    {
+      "epoch": 14.184397163120567,
+      "eval_cer": 0.10924116298709133,
+      "eval_loss": 0.18963314592838287,
+      "eval_runtime": 20.9261,
+      "eval_samples_per_second": 23.894,
+      "eval_steps_per_second": 3.011,
+      "eval_wer": 0.998,
+      "step": 2000
+    },
+    {
+      "epoch": 14.893617021276595,
+      "eval_cer": 0.10945228616238388,
+      "eval_loss": 0.1838211864233017,
+      "eval_runtime": 20.7677,
+      "eval_samples_per_second": 24.076,
+      "eval_steps_per_second": 3.034,
+      "eval_wer": 1.0,
+      "step": 2100
+    },
+    {
+      "epoch": 15.602836879432624,
+      "eval_cer": 0.10806490529617566,
+      "eval_loss": 0.17658106982707977,
+      "eval_runtime": 21.0918,
+      "eval_samples_per_second": 23.706,
+      "eval_steps_per_second": 2.987,
+      "eval_wer": 1.0,
+      "step": 2200
+    },
+    {
+      "epoch": 16.31205673758865,
+      "eval_cer": 0.10706961032693932,
+      "eval_loss": 0.1687668263912201,
+      "eval_runtime": 20.9189,
+      "eval_samples_per_second": 23.902,
+      "eval_steps_per_second": 3.012,
+      "eval_wer": 0.998,
+      "step": 2300
+    },
+    {
+      "epoch": 17.02127659574468,
+      "eval_cer": 0.1069188080588732,
+      "eval_loss": 0.16668924689292908,
+      "eval_runtime": 21.205,
+      "eval_samples_per_second": 23.579,
+      "eval_steps_per_second": 2.971,
+      "eval_wer": 0.998,
+      "step": 2400
+    },
+    {
+      "epoch": 17.73049645390071,
+      "grad_norm": 2.1788010597229004,
+      "learning_rate": 5.9903999999999994e-05,
+      "loss": 0.2296,
+      "step": 2500
+    },
+    {
+      "epoch": 17.73049645390071,
+      "eval_cer": 0.10688864760525998,
+      "eval_loss": 0.16431093215942383,
+      "eval_runtime": 21.0484,
+      "eval_samples_per_second": 23.755,
+      "eval_steps_per_second": 2.993,
+      "eval_wer": 1.0,
+      "step": 2500
+    },
+    {
+      "epoch": 18.43971631205674,
+      "eval_cer": 0.10706961032693932,
+      "eval_loss": 0.16017669439315796,
+      "eval_runtime": 20.7713,
+      "eval_samples_per_second": 24.072,
+      "eval_steps_per_second": 3.033,
+      "eval_wer": 1.0,
+      "step": 2600
+    },
+    {
+      "epoch": 19.148936170212767,
+      "eval_cer": 0.10682832669803353,
+      "eval_loss": 0.16536767780780792,
+      "eval_runtime": 21.3386,
+      "eval_samples_per_second": 23.432,
+      "eval_steps_per_second": 2.952,
+      "eval_wer": 1.0,
+      "step": 2700
+    },
+    {
+      "epoch": 19.858156028368793,
+      "eval_cer": 0.10604415490408976,
+      "eval_loss": 0.16166533529758453,
+      "eval_runtime": 21.3062,
+      "eval_samples_per_second": 23.467,
+      "eval_steps_per_second": 2.957,
+      "eval_wer": 0.998,
+      "step": 2800
+    },
+    {
+      "epoch": 20.0,
+      "step": 2820,
+      "total_flos": 4.085369652431226e+18,
+      "train_loss": 2.2413718947282075,
+      "train_runtime": 5214.7646,
+      "train_samples_per_second": 17.259,
+      "train_steps_per_second": 0.541
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 2820,
+  "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": 4.085369652431226e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}