End of training

README.md

@@ -3,6 +3,8 @@ library_name: transformers
 license: apache-2.0
 base_model: facebook/wav2vec2-xls-r-1b
 tags:
+- automatic-speech-recognition
+- bemgen
 - generated_from_trainer
 metrics:
 - wer
@@ -16,10 +18,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # xls-r-1b-bemgen-balanced-model
 
-This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on an unknown dataset.
+This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on the BEMGEN - NA dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3054
-- Wer: 0.4405
+- Loss: 0.2854
+- Wer: 0.4379
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 5.714285714285714,
+    "eval_loss": 0.2854158580303192,
+    "eval_runtime": 20.9534,
+    "eval_samples": 274,
+    "eval_samples_per_second": 13.077,
+    "eval_steps_per_second": 3.293,
+    "eval_wer": 0.4378953421506613,
+    "total_flos": 1.0020885754408006e+19,
+    "train_loss": 2.059260038137436,
+    "train_runtime": 2265.6733,
+    "train_samples": 2238,
+    "train_samples_per_second": 29.634,
+    "train_steps_per_second": 3.708
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 5.714285714285714,
+    "eval_loss": 0.2854158580303192,
+    "eval_runtime": 20.9534,
+    "eval_samples": 274,
+    "eval_samples_per_second": 13.077,
+    "eval_steps_per_second": 3.293,
+    "eval_wer": 0.4378953421506613
+}

runs/Dec26_15-55-02_srvrocgpu011.uct.ac.za/events.out.tfevents.1735223915.srvrocgpu011.uct.ac.za

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64ec3ba3e679d99e0a26fa7151c491138be5ba9d124589210537573d1a6534ae
+size 40

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 5.714285714285714,
+    "total_flos": 1.0020885754408006e+19,
+    "train_loss": 2.059260038137436,
+    "train_runtime": 2265.6733,
+    "train_samples": 2238,
+    "train_samples_per_second": 29.634,
+    "train_steps_per_second": 3.708
+}

trainer_state.json

@@ -0,0 +1,216 @@
+{
+  "best_metric": 0.28544488549232483,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/xls-r-1b-bemgen-balanced-model/checkpoint-1200",
+  "epoch": 5.714285714285714,
+  "eval_steps": 100,
+  "global_step": 1600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.35714285714285715,
+      "eval_loss": 3.739461898803711,
+      "eval_runtime": 21.0209,
+      "eval_samples_per_second": 13.035,
+      "eval_steps_per_second": 3.282,
+      "eval_wer": 1.00747556066705,
+      "step": 100
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "eval_loss": 2.836632013320923,
+      "eval_runtime": 20.8585,
+      "eval_samples_per_second": 13.136,
+      "eval_steps_per_second": 3.308,
+      "eval_wer": 1.0,
+      "step": 200
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "eval_loss": 2.3453426361083984,
+      "eval_runtime": 20.8764,
+      "eval_samples_per_second": 13.125,
+      "eval_steps_per_second": 3.305,
+      "eval_wer": 1.0,
+      "step": 300
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "eval_loss": 0.5265899896621704,
+      "eval_runtime": 21.0105,
+      "eval_samples_per_second": 13.041,
+      "eval_steps_per_second": 3.284,
+      "eval_wer": 0.7475560667050029,
+      "step": 400
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 2.576737880706787,
+      "learning_rate": 3e-05,
+      "loss": 5.5915,
+      "step": 500
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "eval_loss": 0.46810343861579895,
+      "eval_runtime": 21.0369,
+      "eval_samples_per_second": 13.025,
+      "eval_steps_per_second": 3.28,
+      "eval_wer": 0.6963772282921219,
+      "step": 500
+    },
+    {
+      "epoch": 2.142857142857143,
+      "eval_loss": 0.41566550731658936,
+      "eval_runtime": 20.9744,
+      "eval_samples_per_second": 13.064,
+      "eval_steps_per_second": 3.29,
+      "eval_wer": 0.6483611270845313,
+      "step": 600
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 0.3743371367454529,
+      "eval_runtime": 20.9379,
+      "eval_samples_per_second": 13.086,
+      "eval_steps_per_second": 3.295,
+      "eval_wer": 0.5750431282346176,
+      "step": 700
+    },
+    {
+      "epoch": 2.857142857142857,
+      "eval_loss": 0.32862401008605957,
+      "eval_runtime": 21.0408,
+      "eval_samples_per_second": 13.022,
+      "eval_steps_per_second": 3.279,
+      "eval_wer": 0.5146635997699828,
+      "step": 800
+    },
+    {
+      "epoch": 3.2142857142857144,
+      "eval_loss": 0.3155038058757782,
+      "eval_runtime": 21.0256,
+      "eval_samples_per_second": 13.032,
+      "eval_steps_per_second": 3.282,
+      "eval_wer": 0.476710753306498,
+      "step": 900
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 2.65522837638855,
+      "learning_rate": 2.8101265822784812e-05,
+      "loss": 0.5861,
+      "step": 1000
+    },
+    {
+      "epoch": 3.571428571428571,
+      "eval_loss": 0.3056633472442627,
+      "eval_runtime": 21.1354,
+      "eval_samples_per_second": 12.964,
+      "eval_steps_per_second": 3.265,
+      "eval_wer": 0.49108683151236343,
+      "step": 1000
+    },
+    {
+      "epoch": 3.928571428571429,
+      "eval_loss": 0.30086472630500793,
+      "eval_runtime": 20.9686,
+      "eval_samples_per_second": 13.067,
+      "eval_steps_per_second": 3.291,
+      "eval_wer": 0.4611845888441633,
+      "step": 1100
+    },
+    {
+      "epoch": 4.285714285714286,
+      "eval_loss": 0.28544488549232483,
+      "eval_runtime": 20.9958,
+      "eval_samples_per_second": 13.05,
+      "eval_steps_per_second": 3.286,
+      "eval_wer": 0.4381828637147786,
+      "step": 1200
+    },
+    {
+      "epoch": 4.642857142857143,
+      "eval_loss": 0.3015901446342468,
+      "eval_runtime": 21.1404,
+      "eval_samples_per_second": 12.961,
+      "eval_steps_per_second": 3.264,
+      "eval_wer": 0.45169637722829215,
+      "step": 1300
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.29240939021110535,
+      "eval_runtime": 21.1292,
+      "eval_samples_per_second": 12.968,
+      "eval_steps_per_second": 3.266,
+      "eval_wer": 0.44335825186889016,
+      "step": 1400
+    },
+    {
+      "epoch": 5.357142857142857,
+      "grad_norm": 4.473409175872803,
+      "learning_rate": 2.620253164556962e-05,
+      "loss": 0.3557,
+      "step": 1500
+    },
+    {
+      "epoch": 5.357142857142857,
+      "eval_loss": 0.30794966220855713,
+      "eval_runtime": 21.1536,
+      "eval_samples_per_second": 12.953,
+      "eval_steps_per_second": 3.262,
+      "eval_wer": 0.44882116158711904,
+      "step": 1500
+    },
+    {
+      "epoch": 5.714285714285714,
+      "eval_loss": 0.30544233322143555,
+      "eval_runtime": 20.9988,
+      "eval_samples_per_second": 13.048,
+      "eval_steps_per_second": 3.286,
+      "eval_wer": 0.44048303622771706,
+      "step": 1600
+    },
+    {
+      "epoch": 5.714285714285714,
+      "step": 1600,
+      "total_flos": 1.0020885754408006e+19,
+      "train_loss": 2.059260038137436,
+      "train_runtime": 2265.6733,
+      "train_samples_per_second": 29.634,
+      "train_steps_per_second": 3.708
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 8400,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 400,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 4,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 4
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0020885754408006e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}