update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,92 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- en-asr-leaderboard
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
datasets:
|
| 9 |
+
- mn367/radio-test-dataset
|
| 10 |
+
metrics:
|
| 11 |
+
- wer
|
| 12 |
+
model-index:
|
| 13 |
+
- name: Whisper Medium En
|
| 14 |
+
results:
|
| 15 |
+
- task:
|
| 16 |
+
name: Automatic Speech Recognition
|
| 17 |
+
type: automatic-speech-recognition
|
| 18 |
+
dataset:
|
| 19 |
+
name: Radio dataset
|
| 20 |
+
type: mn367/radio-test-dataset
|
| 21 |
+
args: 'config: en, split: test'
|
| 22 |
+
metrics:
|
| 23 |
+
- name: Wer
|
| 24 |
+
type: wer
|
| 25 |
+
value: 30.971856370442257
|
| 26 |
+
---
|
| 27 |
+
|
| 28 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 29 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 30 |
+
|
| 31 |
+
# Whisper Medium En
|
| 32 |
+
|
| 33 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the Radio dataset dataset.
|
| 34 |
+
It achieves the following results on the evaluation set:
|
| 35 |
+
- Loss: 0.6118
|
| 36 |
+
- Wer: 30.9719
|
| 37 |
+
|
| 38 |
+
## Model description
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Intended uses & limitations
|
| 43 |
+
|
| 44 |
+
More information needed
|
| 45 |
+
|
| 46 |
+
## Training and evaluation data
|
| 47 |
+
|
| 48 |
+
More information needed
|
| 49 |
+
|
| 50 |
+
## Training procedure
|
| 51 |
+
|
| 52 |
+
### Training hyperparameters
|
| 53 |
+
|
| 54 |
+
The following hyperparameters were used during training:
|
| 55 |
+
- learning_rate: 1e-06
|
| 56 |
+
- train_batch_size: 16
|
| 57 |
+
- eval_batch_size: 8
|
| 58 |
+
- seed: 42
|
| 59 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 60 |
+
- lr_scheduler_type: linear
|
| 61 |
+
- lr_scheduler_warmup_steps: 500
|
| 62 |
+
- training_steps: 1600
|
| 63 |
+
- mixed_precision_training: Native AMP
|
| 64 |
+
|
| 65 |
+
### Training results
|
| 66 |
+
|
| 67 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 68 |
+
|:-------------:|:-----:|:----:|:---------------:|:-------:|
|
| 69 |
+
| 3.6505 | 0.75 | 100 | 3.5819 | 58.0618 |
|
| 70 |
+
| 2.7405 | 1.5 | 200 | 2.5030 | 47.7471 |
|
| 71 |
+
| 1.6934 | 2.26 | 300 | 1.6058 | 36.4342 |
|
| 72 |
+
| 0.8876 | 3.01 | 400 | 0.7653 | 39.4843 |
|
| 73 |
+
| 0.6136 | 3.76 | 500 | 0.6374 | 35.2281 |
|
| 74 |
+
| 0.4993 | 4.51 | 600 | 0.5950 | 34.9785 |
|
| 75 |
+
| 0.3364 | 5.26 | 700 | 0.5776 | 28.6150 |
|
| 76 |
+
| 0.4268 | 6.02 | 800 | 0.5640 | 29.7241 |
|
| 77 |
+
| 0.3367 | 6.77 | 900 | 0.5706 | 29.8489 |
|
| 78 |
+
| 0.2706 | 7.52 | 1000 | 0.5757 | 28.0189 |
|
| 79 |
+
| 0.2932 | 8.27 | 1100 | 0.5789 | 28.3377 |
|
| 80 |
+
| 0.2699 | 9.02 | 1200 | 0.5869 | 29.9875 |
|
| 81 |
+
| 0.2434 | 9.77 | 1300 | 0.5976 | 30.1816 |
|
| 82 |
+
| 0.267 | 10.53 | 1400 | 0.6020 | 30.5559 |
|
| 83 |
+
| 0.2364 | 11.28 | 1500 | 0.6106 | 30.6391 |
|
| 84 |
+
| 0.2095 | 12.03 | 1600 | 0.6118 | 30.9719 |
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
### Framework versions
|
| 88 |
+
|
| 89 |
+
- Transformers 4.25.0.dev0
|
| 90 |
+
- Pytorch 1.12.1+cu113
|
| 91 |
+
- Datasets 2.7.0
|
| 92 |
+
- Tokenizers 0.13.2
|