Update README.md
Browse files
README.md
CHANGED
|
@@ -18,6 +18,41 @@ grammatical mistakes, slang, and non-native speaker errors. This model helps imp
|
|
| 18 |
in scenarios where speakers use incorrect or informal English, making it useful in language learning,
|
| 19 |
transcription of casual conversations, or analyzing spoken communication from non-native English speakers.
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
## Usage Guide
|
| 22 |
|
| 23 |
This project was executed on an Ubuntu 22.04.3 system running Linux kernel 6.8.0-40-generic.
|
|
@@ -81,5 +116,12 @@ tokenizer = WhisperTokenizer.from_pretrained("openai/whisper-large-v3", task="tr
|
|
| 81 |
feature_extractor = WhisperFeatureExtractor.from_pretrained("openai/whisper-large-v3")
|
| 82 |
|
| 83 |
pipe = pipeline(model=model, tokenizer=tokenizer, feature_extractor=feature_extractor, task="automatic-speech-recognition", device=device)
|
| 84 |
-
```
|
| 85 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
in scenarios where speakers use incorrect or informal English, making it useful in language learning,
|
| 19 |
transcription of casual conversations, or analyzing spoken communication from non-native English speakers.
|
| 20 |
|
| 21 |
+
## Training procedure
|
| 22 |
+
|
| 23 |
+
### Training hyperparameters
|
| 24 |
+
|
| 25 |
+
The following hyperparameters were used during training:
|
| 26 |
+
- learning_rate: 1e-05
|
| 27 |
+
- train_batch_size: 8
|
| 28 |
+
- eval_batch_size: 8
|
| 29 |
+
- seed: 42
|
| 30 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 31 |
+
- lr_scheduler_type: linear
|
| 32 |
+
- lr_scheduler_warmup_steps: 50
|
| 33 |
+
- training_steps: 100000
|
| 34 |
+
- mixed_precision_training: Native AMP
|
| 35 |
+
|
| 36 |
+
### Training results
|
| 37 |
+
|
| 38 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 39 |
+
|:-------------:|:------:|:----:|:---------------:|:-------:|
|
| 40 |
+
| 0.9094 | 0.1270 | 500 | 0.6347 | 24.3686 |
|
| 41 |
+
| 0.5517 | 0.2541 | 1000 | 0.4835 | 18.0769 |
|
| 42 |
+
| 0.5364 | 0.3811 | 1500 | 0.4330 | 15.1149 |
|
| 43 |
+
| 0.5503 | 0.5081 | 2000 | 0.4113 | 13.6524 |
|
| 44 |
+
| 0.6521 | 0.6352 | 2500 | 0.3987 | 13.5897 |
|
| 45 |
+
| 0.6044 | 0.7622 | 3000 | 0.3912 | 13.0538 |
|
| 46 |
+
| 0.5487 | 0.8892 | 3500 | 0.3835 | 12.6119 |
|
| 47 |
+
| 0.5297 | 1.0163 | 4000 | 0.3791 | 12.4408 |
|
| 48 |
+
| 0.46 | 1.1433 | 4500 | 0.3751 | 12.3525 |
|
| 49 |
+
| 0.4947 | 1.2703 | 5000 | 0.3721 | 12.1415 |
|
| 50 |
+
| 0.524 | 1.3974 | 5500 | 0.3682 | 13.0139 |
|
| 51 |
+
| 0.4743 | 1.5244 | 6000 | 0.3649 | 13.3388 |
|
| 52 |
+
| 0.5338 | 1.6514 | 6500 | 0.3621 | 12.9397 |
|
| 53 |
+
| 0.5162 | 1.7785 | 7000 | 0.3597 | 13.3246 |
|
| 54 |
+
| 0.5004 | 1.9055 | 7500 | 0.3590 | 12.3268 |
|
| 55 |
+
|
| 56 |
## Usage Guide
|
| 57 |
|
| 58 |
This project was executed on an Ubuntu 22.04.3 system running Linux kernel 6.8.0-40-generic.
|
|
|
|
| 116 |
feature_extractor = WhisperFeatureExtractor.from_pretrained("openai/whisper-large-v3")
|
| 117 |
|
| 118 |
pipe = pipeline(model=model, tokenizer=tokenizer, feature_extractor=feature_extractor, task="automatic-speech-recognition", device=device)
|
|
|
|
| 119 |
|
| 120 |
+
|
| 121 |
+
### Framework versions
|
| 122 |
+
|
| 123 |
+
- PEFT 0.11.1
|
| 124 |
+
- Transformers 4.42.4
|
| 125 |
+
- Pytorch 2.1.0+cu118
|
| 126 |
+
- Datasets 2.20.0
|
| 127 |
+
- Tokenizers 0.19.1
|