update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,81 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: MilladRN
|
| 7 |
+
results: []
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 11 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 12 |
+
|
| 13 |
+
# MilladRN
|
| 14 |
+
|
| 15 |
+
This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
|
| 16 |
+
It achieves the following results on the evaluation set:
|
| 17 |
+
- Loss: 3.4355
|
| 18 |
+
- Wer: 0.4907
|
| 19 |
+
- Cer: 0.2802
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 0.0001
|
| 39 |
+
- train_batch_size: 8
|
| 40 |
+
- eval_batch_size: 8
|
| 41 |
+
- seed: 42
|
| 42 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 43 |
+
- lr_scheduler_type: linear
|
| 44 |
+
- lr_scheduler_warmup_steps: 4000
|
| 45 |
+
- num_epochs: 750
|
| 46 |
+
- mixed_precision_training: Native AMP
|
| 47 |
+
|
| 48 |
+
### Training results
|
| 49 |
+
|
| 50 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 51 |
+
|:-------------:|:------:|:-----:|:---------------:|:------:|:------:|
|
| 52 |
+
| 3.3347 | 33.9 | 2000 | 2.2561 | 0.9888 | 0.6087 |
|
| 53 |
+
| 1.3337 | 67.8 | 4000 | 1.8137 | 0.6877 | 0.3407 |
|
| 54 |
+
| 0.6504 | 101.69 | 6000 | 2.0718 | 0.6245 | 0.3229 |
|
| 55 |
+
| 0.404 | 135.59 | 8000 | 2.2246 | 0.6004 | 0.3221 |
|
| 56 |
+
| 0.2877 | 169.49 | 10000 | 2.2624 | 0.5836 | 0.3107 |
|
| 57 |
+
| 0.2149 | 203.39 | 12000 | 2.3788 | 0.5279 | 0.2802 |
|
| 58 |
+
| 0.1693 | 237.29 | 14000 | 1.8928 | 0.5502 | 0.2937 |
|
| 59 |
+
| 0.1383 | 271.19 | 16000 | 2.7520 | 0.5725 | 0.3103 |
|
| 60 |
+
| 0.1169 | 305.08 | 18000 | 2.2552 | 0.5446 | 0.2968 |
|
| 61 |
+
| 0.1011 | 338.98 | 20000 | 2.6794 | 0.5725 | 0.3119 |
|
| 62 |
+
| 0.0996 | 372.88 | 22000 | 2.4704 | 0.5595 | 0.3142 |
|
| 63 |
+
| 0.0665 | 406.78 | 24000 | 2.9073 | 0.5836 | 0.3194 |
|
| 64 |
+
| 0.0538 | 440.68 | 26000 | 3.1357 | 0.5632 | 0.3213 |
|
| 65 |
+
| 0.0538 | 474.58 | 28000 | 2.5639 | 0.5613 | 0.3091 |
|
| 66 |
+
| 0.0493 | 508.47 | 30000 | 3.3801 | 0.5613 | 0.3119 |
|
| 67 |
+
| 0.0451 | 542.37 | 32000 | 3.5469 | 0.5428 | 0.3158 |
|
| 68 |
+
| 0.0307 | 576.27 | 34000 | 4.2243 | 0.5390 | 0.3126 |
|
| 69 |
+
| 0.0301 | 610.17 | 36000 | 3.6666 | 0.5297 | 0.2929 |
|
| 70 |
+
| 0.0269 | 644.07 | 38000 | 3.2164 | 0.5 | 0.2838 |
|
| 71 |
+
| 0.0182 | 677.97 | 40000 | 3.0557 | 0.4963 | 0.2779 |
|
| 72 |
+
| 0.0191 | 711.86 | 42000 | 3.5190 | 0.5130 | 0.2921 |
|
| 73 |
+
| 0.0133 | 745.76 | 44000 | 3.4355 | 0.4907 | 0.2802 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
### Framework versions
|
| 77 |
+
|
| 78 |
+
- Transformers 4.17.0
|
| 79 |
+
- Pytorch 1.12.0+cu113
|
| 80 |
+
- Datasets 1.18.3
|
| 81 |
+
- Tokenizers 0.12.1
|