docs: advanced model card (Armenian SpeechT5)
Browse files
README.md
CHANGED
|
@@ -1,71 +1,25 @@
|
|
| 1 |
---
|
| 2 |
-
|
|
|
|
|
|
|
| 3 |
tags:
|
| 4 |
-
-
|
| 5 |
-
|
| 6 |
-
-
|
| 7 |
-
results: []
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
| 11 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 12 |
|
| 13 |
-
|
| 14 |
|
| 15 |
-
|
| 16 |
-
It achieves the following results on the evaluation set:
|
| 17 |
-
- Loss: 0.4601
|
| 18 |
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
-
|
| 22 |
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
More information needed
|
| 26 |
-
|
| 27 |
-
## Training and evaluation data
|
| 28 |
-
|
| 29 |
-
More information needed
|
| 30 |
-
|
| 31 |
-
## Training procedure
|
| 32 |
-
|
| 33 |
-
### Training hyperparameters
|
| 34 |
-
|
| 35 |
-
The following hyperparameters were used during training:
|
| 36 |
-
- learning_rate: 2e-05
|
| 37 |
-
- train_batch_size: 4
|
| 38 |
-
- eval_batch_size: 2
|
| 39 |
-
- seed: 42
|
| 40 |
-
- gradient_accumulation_steps: 8
|
| 41 |
-
- total_train_batch_size: 32
|
| 42 |
-
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 43 |
-
- lr_scheduler_type: linear
|
| 44 |
-
- lr_scheduler_warmup_steps: 1000
|
| 45 |
-
- training_steps: 6000
|
| 46 |
-
- mixed_precision_training: Native AMP
|
| 47 |
-
|
| 48 |
-
### Training results
|
| 49 |
-
|
| 50 |
-
| Training Loss | Epoch | Step | Validation Loss |
|
| 51 |
-
|:-------------:|:--------:|:----:|:---------------:|
|
| 52 |
-
| 0.6192 | 27.3973 | 500 | 0.5595 |
|
| 53 |
-
| 0.5686 | 54.7945 | 1000 | 0.5097 |
|
| 54 |
-
| 0.537 | 82.1918 | 1500 | 0.4940 |
|
| 55 |
-
| 0.5239 | 109.5890 | 2000 | 0.4815 |
|
| 56 |
-
| 0.5143 | 136.9863 | 2500 | 0.4775 |
|
| 57 |
-
| 0.5087 | 166.2740 | 3000 | 0.4622 |
|
| 58 |
-
| 0.5007 | 193.6712 | 3500 | 0.4592 |
|
| 59 |
-
| 0.5083 | 137.5527 | 4000 | 0.4671 |
|
| 60 |
-
| 0.5025 | 154.4304 | 4500 | 0.4657 |
|
| 61 |
-
| 0.4963 | 171.3080 | 5000 | 0.4626 |
|
| 62 |
-
| 0.4982 | 188.1857 | 5500 | 0.4603 |
|
| 63 |
-
| 0.4939 | 205.0633 | 6000 | 0.4601 |
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
### Framework versions
|
| 67 |
-
|
| 68 |
-
- Transformers 4.43.3
|
| 69 |
-
- Pytorch 2.4.0+cu121
|
| 70 |
-
- Datasets 2.20.0
|
| 71 |
-
- Tokenizers 0.19.1
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- hy
|
| 4 |
+
license: mit
|
| 5 |
tags:
|
| 6 |
+
- deprecated
|
| 7 |
+
- speecht5
|
| 8 |
+
- armenian
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
+
# ⚠️ Deprecated: `TTS_NB`
|
|
|
|
| 12 |
|
| 13 |
+
This repository is **deprecated** and scheduled for removal.
|
| 14 |
|
| 15 |
+
## Use instead
|
|
|
|
|
|
|
| 16 |
|
| 17 |
+
| Purpose | Replacement |
|
| 18 |
+
|---------|-------------|
|
| 19 |
+
| Training / fine-tuning | [Edmon02/TTS_NB_2](https://huggingface.co/Edmon02/TTS_NB_2) |
|
| 20 |
+
| Inference (PyTorch) | [Edmon02/speecht5_finetuned_voxpopuli_hy](https://huggingface.co/Edmon02/speecht5_finetuned_voxpopuli_hy) |
|
| 21 |
+
| ONNX | [Edmon02/TTS_NB_ONNX](https://huggingface.co/Edmon02/TTS_NB_ONNX) |
|
| 22 |
|
| 23 |
+
**Reason:** Intermediate checkpoint with incorrect `base_model` metadata, superseded by `TTS_NB_2` (vocab 158).
|
| 24 |
|
| 25 |
+
Do not start new projects on this repo.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|