End of training

README.md

@@ -3,6 +3,8 @@ library_name: transformers
 license: mit
 base_model: facebook/w2v-bert-2.0
 tags:
+- automatic-speech-recognition
+- genbed
 - generated_from_trainer
 metrics:
 - wer
@@ -16,10 +18,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # w2v-bert-bem-genbed-m-model
 
-This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on an unknown dataset.
+This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the GENBED - BEM dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4253
-- Wer: 0.5296
+- Loss: 0.4168
+- Wer: 0.5478
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 11.019283746556473,
+    "eval_loss": 0.4167534410953522,
+    "eval_runtime": 40.9232,
+    "eval_samples": 969,
+    "eval_samples_per_second": 23.679,
+    "eval_steps_per_second": 2.981,
+    "eval_wer": 0.5478185181155478,
+    "total_flos": 5.988611611546594e+18,
+    "train_loss": 0.38784495496749877,
+    "train_runtime": 3222.307,
+    "train_samples": 2900,
+    "train_samples_per_second": 26.999,
+    "train_steps_per_second": 1.685
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 11.019283746556473,
+    "eval_loss": 0.4167534410953522,
+    "eval_runtime": 40.9232,
+    "eval_samples": 969,
+    "eval_samples_per_second": 23.679,
+    "eval_steps_per_second": 2.981,
+    "eval_wer": 0.5478185181155478
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 11.019283746556473,
+    "total_flos": 5.988611611546594e+18,
+    "train_loss": 0.38784495496749877,
+    "train_runtime": 3222.307,
+    "train_samples": 2900,
+    "train_samples_per_second": 26.999,
+    "train_steps_per_second": 1.685
+}

trainer_state.json

@@ -0,0 +1,281 @@
+{
+  "best_metric": 0.4167534410953522,
+  "best_model_checkpoint": "/scratch/skscla001/results/w2v-bert-bem-genbed-m-model/checkpoint-1400",
+  "epoch": 11.019283746556473,
+  "eval_steps": 200,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5509641873278237,
+      "grad_norm": 2.760388135910034,
+      "learning_rate": 0.000294,
+      "loss": 1.5066,
+      "step": 100
+    },
+    {
+      "epoch": 1.1019283746556474,
+      "grad_norm": 1.6717568635940552,
+      "learning_rate": 0.0002945403377110694,
+      "loss": 0.7215,
+      "step": 200
+    },
+    {
+      "epoch": 1.1019283746556474,
+      "eval_loss": 0.6149704456329346,
+      "eval_runtime": 41.354,
+      "eval_samples_per_second": 23.432,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 0.743009465781743,
+      "step": 200
+    },
+    {
+      "epoch": 1.6528925619834711,
+      "grad_norm": 1.9287854433059692,
+      "learning_rate": 0.00028891181988742964,
+      "loss": 0.597,
+      "step": 300
+    },
+    {
+      "epoch": 2.203856749311295,
+      "grad_norm": 1.757243275642395,
+      "learning_rate": 0.00028328330206378984,
+      "loss": 0.5519,
+      "step": 400
+    },
+    {
+      "epoch": 2.203856749311295,
+      "eval_loss": 0.5605354309082031,
+      "eval_runtime": 41.456,
+      "eval_samples_per_second": 23.374,
+      "eval_steps_per_second": 2.943,
+      "eval_wer": 0.7115656620607116,
+      "step": 400
+    },
+    {
+      "epoch": 2.7548209366391183,
+      "grad_norm": 1.5606329441070557,
+      "learning_rate": 0.0002776547842401501,
+      "loss": 0.485,
+      "step": 500
+    },
+    {
+      "epoch": 3.3057851239669422,
+      "grad_norm": 0.9466643333435059,
+      "learning_rate": 0.0002720262664165103,
+      "loss": 0.4346,
+      "step": 600
+    },
+    {
+      "epoch": 3.3057851239669422,
+      "eval_loss": 0.4709364175796509,
+      "eval_runtime": 41.6444,
+      "eval_samples_per_second": 23.268,
+      "eval_steps_per_second": 2.93,
+      "eval_wer": 0.6377978457186378,
+      "step": 600
+    },
+    {
+      "epoch": 3.8567493112947657,
+      "grad_norm": 0.8296416997909546,
+      "learning_rate": 0.0002663977485928705,
+      "loss": 0.3911,
+      "step": 700
+    },
+    {
+      "epoch": 4.40771349862259,
+      "grad_norm": 1.098000407218933,
+      "learning_rate": 0.00026076923076923076,
+      "loss": 0.3545,
+      "step": 800
+    },
+    {
+      "epoch": 4.40771349862259,
+      "eval_loss": 0.4685700237751007,
+      "eval_runtime": 41.6691,
+      "eval_samples_per_second": 23.255,
+      "eval_steps_per_second": 2.928,
+      "eval_wer": 0.5984114895005984,
+      "step": 800
+    },
+    {
+      "epoch": 4.958677685950414,
+      "grad_norm": 0.8255869746208191,
+      "learning_rate": 0.00025514071294559096,
+      "loss": 0.3439,
+      "step": 900
+    },
+    {
+      "epoch": 5.509641873278237,
+      "grad_norm": 0.8485454320907593,
+      "learning_rate": 0.00024951219512195117,
+      "loss": 0.3004,
+      "step": 1000
+    },
+    {
+      "epoch": 5.509641873278237,
+      "eval_loss": 0.4578340947628021,
+      "eval_runtime": 41.6261,
+      "eval_samples_per_second": 23.279,
+      "eval_steps_per_second": 2.931,
+      "eval_wer": 0.6202807093896203,
+      "step": 1000
+    },
+    {
+      "epoch": 6.0606060606060606,
+      "grad_norm": 0.5917987823486328,
+      "learning_rate": 0.00024388367729831143,
+      "loss": 0.2984,
+      "step": 1100
+    },
+    {
+      "epoch": 6.6115702479338845,
+      "grad_norm": 0.8203296065330505,
+      "learning_rate": 0.00023825515947467166,
+      "loss": 0.2498,
+      "step": 1200
+    },
+    {
+      "epoch": 6.6115702479338845,
+      "eval_loss": 0.4245435893535614,
+      "eval_runtime": 41.2956,
+      "eval_samples_per_second": 23.465,
+      "eval_steps_per_second": 2.954,
+      "eval_wer": 0.5246436731585247,
+      "step": 1200
+    },
+    {
+      "epoch": 7.162534435261708,
+      "grad_norm": 0.9434472322463989,
+      "learning_rate": 0.0002326266416510319,
+      "loss": 0.2456,
+      "step": 1300
+    },
+    {
+      "epoch": 7.7134986225895315,
+      "grad_norm": 1.4402457475662231,
+      "learning_rate": 0.0002269981238273921,
+      "loss": 0.23,
+      "step": 1400
+    },
+    {
+      "epoch": 7.7134986225895315,
+      "eval_loss": 0.4167534410953522,
+      "eval_runtime": 41.5017,
+      "eval_samples_per_second": 23.348,
+      "eval_steps_per_second": 2.94,
+      "eval_wer": 0.5478185181155478,
+      "step": 1400
+    },
+    {
+      "epoch": 8.264462809917354,
+      "grad_norm": 0.6520706415176392,
+      "learning_rate": 0.00022136960600375232,
+      "loss": 0.2146,
+      "step": 1500
+    },
+    {
+      "epoch": 8.81542699724518,
+      "grad_norm": 0.4538098871707916,
+      "learning_rate": 0.00021574108818011255,
+      "loss": 0.1959,
+      "step": 1600
+    },
+    {
+      "epoch": 8.81542699724518,
+      "eval_loss": 0.4212118089199066,
+      "eval_runtime": 41.8307,
+      "eval_samples_per_second": 23.165,
+      "eval_steps_per_second": 2.917,
+      "eval_wer": 0.523011641823523,
+      "step": 1600
+    },
+    {
+      "epoch": 9.366391184573002,
+      "grad_norm": 0.5852454900741577,
+      "learning_rate": 0.00021011257035647278,
+      "loss": 0.1855,
+      "step": 1700
+    },
+    {
+      "epoch": 9.917355371900827,
+      "grad_norm": 0.6804907321929932,
+      "learning_rate": 0.00020448405253283299,
+      "loss": 0.1682,
+      "step": 1800
+    },
+    {
+      "epoch": 9.917355371900827,
+      "eval_loss": 0.43570414185523987,
+      "eval_runtime": 41.3416,
+      "eval_samples_per_second": 23.439,
+      "eval_steps_per_second": 2.951,
+      "eval_wer": 0.5053857034055054,
+      "step": 1800
+    },
+    {
+      "epoch": 10.46831955922865,
+      "grad_norm": 0.7718915939331055,
+      "learning_rate": 0.00019885553470919322,
+      "loss": 0.1367,
+      "step": 1900
+    },
+    {
+      "epoch": 11.019283746556473,
+      "grad_norm": 0.7347743511199951,
+      "learning_rate": 0.00019322701688555345,
+      "loss": 0.1459,
+      "step": 2000
+    },
+    {
+      "epoch": 11.019283746556473,
+      "eval_loss": 0.4253489673137665,
+      "eval_runtime": 42.1972,
+      "eval_samples_per_second": 22.964,
+      "eval_steps_per_second": 2.891,
+      "eval_wer": 0.5296485692525297,
+      "step": 2000
+    },
+    {
+      "epoch": 11.019283746556473,
+      "step": 2000,
+      "total_flos": 5.988611611546594e+18,
+      "train_loss": 0.38784495496749877,
+      "train_runtime": 3222.307,
+      "train_samples_per_second": 26.999,
+      "train_steps_per_second": 1.685
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 5430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 3
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.988611611546594e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}