End of training

README.md

@@ -2,7 +2,7 @@
 language:
 - nl
 license: apache-2.0
-base_model: openai/whisper-small
+base_model: openai/whisper-medium
 tags:
 - nl-asr-leaderboard
 - generated_from_trainer
@@ -11,7 +11,7 @@ datasets:
 metrics:
 - wer
 model-index:
-- name: Whisper Small NL - Noise
+- name: Whisper Medium NL - Noise
   results:
   - task:
       name: Automatic Speech Recognition
@@ -20,23 +20,23 @@ model-index:
       name: Common Voice 11.0
       type: mozilla-foundation/common_voice_11_0
       config: nl
-      split: None
+      split: test[:250]
       args: 'config: nl, split: test'
     metrics:
     - name: Wer
       type: wer
-      value: 32.565492321589886
+      value: 22.04155374887082
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# Whisper Small NL - Noise
+# Whisper Medium NL - Noise
 
-This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 11.0 dataset.
+This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the Common Voice 11.0 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.5674
-- Wer: 32.5655
+- Loss: 0.3982
+- Wer: 22.0416
 
 ## Model description
 
@@ -62,15 +62,15 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 500
-- training_steps: 4000
+- training_steps: 10000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Wer     |
-|:-------------:|:-----:|:----:|:---------------:|:-------:|
-| 0.3571        | 0.39  | 2000 | 0.6279          | 35.4562 |
-| 0.2991        | 0.78  | 4000 | 0.5674          | 32.5655 |
+| Training Loss | Epoch | Step  | Validation Loss | Wer     |
+|:-------------:|:-----:|:-----:|:---------------:|:-------:|
+| 0.1974        | 0.97  | 5000  | 0.4473          | 25.7904 |
+| 0.0668        | 1.95  | 10000 | 0.3982          | 22.0416 |
 
 
 ### Framework versions

generation_config.json

@@ -1,44 +1,28 @@
 {
   "alignment_heads": [
     [
-      5,
-      3
+      13,
+      15
     ],
     [
-      5,
-      9
-    ],
-    [
-      8,
-      0
-    ],
-    [
-      8,
+      15,
       4
     ],
     [
-      8,
-      7
+      15,
+      15
     ],
     [
-      8,
-      8
+      16,
+      1
     ],
     [
-      9,
+      20,
       0
     ],
     [
-      9,
-      7
-    ],
-    [
-      9,
-      9
-    ],
-    [
-      10,
-      5
+      23,
+      4
     ]
   ],
   "begin_suppress_tokens": [