Merge branch 'main' of https://huggingface.co/patrickvonplaten/wav2vec2-large-xls-r-300m-common_voice-tr-ft into main

README.md

@@ -20,9 +20,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the COMMON_VOICE - TR dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3823
-- Wer: 0.3209
-- Cer: 0.0771
+- Loss: 0.4179
+- Wer: 0.3071
+- Cer: 0.0736
 
 ## Model description
 
@@ -52,7 +52,7 @@ 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
-- num_epochs: 75.0
+- num_epochs: 100.0
 - mixed_precision_training: Native AMP
 
 ### Training results
@@ -64,9 +64,11 @@ The following hyperparameters were used during training:
 | 0.4343        | 27.27 | 1500 | 0.3819          | 0.4060 | 0.1012 |
 | 0.4401        | 36.36 | 2000 | 0.3991          | 0.3954 | 0.1001 |
 | 0.2647        | 45.45 | 2500 | 0.3901          | 0.3689 | 0.0914 |
-| 0.2656        | 54.55 | 3000 | 0.3866          | 0.3463 | 0.0852 |
-| 0.2586        | 63.64 | 3500 | 0.3779          | 0.3297 | 0.0804 |
-| 0.2041        | 72.73 | 4000 | 0.3854          | 0.3234 | 0.0776 |
+| 0.2656        | 54.55 | 3000 | 0.4284          | 0.3463 | 0.0852 |
+| 0.2586        | 63.64 | 3500 | 0.4084          | 0.3297 | 0.0804 |
+| 0.2041        | 72.73 | 4000 | 0.3907          | 0.3193 | 0.0781 |
+| 0.4265        | 81.82 | 4500 | 0.4265          | 0.3120 | 0.0755 |
+| 0.2041        | 90.91 | 5000 | 0.4240          | 0.3071 | 0.0736 |
 
 
 ### Framework versions