End of training

README.md

@@ -16,6 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
 # speecht5_tts_sinhala
 
 This model is a fine-tuned version of [microsoft/speecht5_tts](https://huggingface.co/microsoft/speecht5_tts) on the sinhala_tts dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.4122
 
 ## Model description
 
@@ -35,15 +37,26 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
-- train_batch_size: 64
+- train_batch_size: 16
 - eval_batch_size: 8
 - seed: 42
 - gradient_accumulation_steps: 2
-- total_train_batch_size: 128
+- total_train_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 500
 - training_steps: 4000
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.4719        | 28.17  | 1000 | 0.4329          |
+| 0.4419        | 56.34  | 2000 | 0.4149          |
+| 0.4281        | 84.51  | 3000 | 0.4132          |
+| 0.4358        | 112.68 | 4000 | 0.4122          |
+
 
 ### Framework versions