End of training

README.md

@@ -1,40 +1,42 @@
 ---
+language:
+- dv
 license: apache-2.0
-base_model: openai/whisper-tiny
+base_model: openai/whisper-small
 tags:
 - generated_from_trainer
 datasets:
-- PolyAI/minds14
+- mozilla-foundation/common_voice_13_0
 metrics:
 - wer
 model-index:
-- name: whisper-small-dv
+- name: Whisper-Small-Dv-fine-tuned
   results:
   - task:
       name: Automatic Speech Recognition
       type: automatic-speech-recognition
     dataset:
-      name: PolyAI/minds14
-      type: PolyAI/minds14
-      config: en-US
-      split: train
-      args: en-US
+      name: Common Voice 13
+      type: mozilla-foundation/common_voice_13_0
+      config: dv
+      split: test
+      args: dv
     metrics:
     - name: Wer
       type: wer
-      value: 35.6091030789826
+      value: 12.868171227874953
 ---
 
 <!-- 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-dv
+# Whisper-Small-Dv-fine-tuned
 
-This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the PolyAI/minds14 dataset.
+This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 13 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7160
-- Wer Ortho: 36.2369
-- Wer: 35.6091
+- Loss: 0.1666
+- Wer Ortho: 60.2479
+- Wer: 12.8682
 
 ## Model description
 
@@ -62,15 +64,15 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: constant_with_warmup
 - lr_scheduler_warmup_steps: 50
-- training_steps: 200
+- training_steps: 500
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch   | Step | Validation Loss | Wer Ortho | Wer     |
-|:-------------:|:-------:|:----:|:---------------:|:---------:|:-------:|
-| 0.2296        | 7.1429  | 100  | 0.5760          | 34.1463   | 33.7349 |
-| 0.0048        | 14.2857 | 200  | 0.7160          | 36.2369   | 35.6091 |
+| Training Loss | Epoch  | Step | Validation Loss | Wer Ortho | Wer     |
+|:-------------:|:------:|:----:|:---------------:|:---------:|:-------:|
+| 0.1858        | 1.6313 | 250  | 0.2034          | 69.6915   | 15.8101 |
+| 0.0747        | 3.2626 | 500  | 0.1666          | 60.2479   | 12.8682 |
 
 
 ### Framework versions

generation_config.json

@@ -1,27 +1,43 @@
 {
   "alignment_heads": [
     [
-      2,
-      2
+      5,
+      3
+    ],
+    [
+      5,
+      9
     ],
     [
-      3,
+      8,
       0
     ],
     [
-      3,
-      2
+      8,
+      4
     ],
     [
-      3,
-      3
+      8,
+      7
     ],
     [
-      3,
-      4
+      8,
+      8
+    ],
+    [
+      9,
+      0
+    ],
+    [
+      9,
+      7
+    ],
+    [
+      9,
+      9
     ],
     [
-      3,
+      10,
       5
     ]
   ],