mikhail-panzo
/

zlm_b128_le5_s8000

@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [microsoft/speecht5_tts](https://huggingface.co/microsoft/speecht5_tts) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3566
 ## Model description
@@ -35,10 +35,10 @@ More information needed
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
-- train_batch_size: 32
 - eval_batch_size: 8
 - seed: 42
-- gradient_accumulation_steps: 4
 - total_train_batch_size: 128
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
@@ -50,22 +50,22 @@ The following hyperparameters were used during training:
 | Training Loss | Epoch   | Step | Validation Loss |
 |:-------------:|:-------:|:----:|:---------------:|
-| 0.4378        | 0.8375  | 500  | 0.3956          |
-| 0.4328        | 1.6750  | 1000 | 0.3922          |
-| 0.4228        | 2.5126  | 1500 | 0.3871          |
-| 0.4179        | 3.3501  | 2000 | 0.3843          |
-| 0.409         | 4.1876  | 2500 | 0.3769          |
-| 0.41          | 5.0251  | 3000 | 0.3739          |
-| 0.4003        | 5.8626  | 3500 | 0.3709          |
-| 0.4052        | 6.7002  | 4000 | 0.3680          |
-| 0.397         | 7.5377  | 4500 | 0.3636          |
-| 0.4007        | 8.3752  | 5000 | 0.3615          |
-| 0.4022        | 9.2127  | 5500 | 0.3612          |
-| 0.3896        | 10.0503 | 6000 | 0.3601          |
-| 0.3911        | 10.8878 | 6500 | 0.3577          |
-| 0.3949        | 11.7253 | 7000 | 0.3584          |
-| 0.3948        | 12.5628 | 7500 | 0.3568          |
-| 0.3877        | 13.4003 | 8000 | 0.3566          |
 ### Framework versions

 This model is a fine-tuned version of [microsoft/speecht5_tts](https://huggingface.co/microsoft/speecht5_tts) on an unknown dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.3662
 ## Model description
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
+- train_batch_size: 16
 - eval_batch_size: 8
 - seed: 42
+- gradient_accumulation_steps: 8
 - total_train_batch_size: 128
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 | Training Loss | Epoch   | Step | Validation Loss |
 |:-------------:|:-------:|:----:|:---------------:|
+| 0.6645        | 0.8377  | 500  | 0.5698          |
+| 0.5581        | 1.6754  | 1000 | 0.4794          |
+| 0.5045        | 2.5131  | 1500 | 0.4467          |
+| 0.4776        | 3.3508  | 2000 | 0.4236          |
+| 0.4553        | 4.1885  | 2500 | 0.4093          |
+| 0.4489        | 5.0262  | 3000 | 0.3968          |
+| 0.4337        | 5.8639  | 3500 | 0.3926          |
+| 0.4282        | 6.7016  | 4000 | 0.3837          |
+| 0.4188        | 7.5393  | 4500 | 0.3798          |
+| 0.4222        | 8.3770  | 5000 | 0.3784          |
+| 0.412         | 9.2147  | 5500 | 0.3729          |
+| 0.4056        | 10.0524 | 6000 | 0.3697          |
+| 0.4065        | 10.8901 | 6500 | 0.3685          |
+| 0.4069        | 11.7277 | 7000 | 0.3675          |
+| 0.4049        | 12.5654 | 7500 | 0.3666          |
+| 0.4044        | 13.4031 | 8000 | 0.3662          |
 ### Framework versions