Commit
•
c9e6513
1
Parent(s):
166f720
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- haseong8012/child-50k
|
| 5 |
+
language:
|
| 6 |
+
- ko
|
| 7 |
+
metrics:
|
| 8 |
+
- wer
|
| 9 |
+
- cer
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 13 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 14 |
+
|
| 15 |
+
# whisper-small_child50K_cosLR
|
| 16 |
+
|
| 17 |
+
This model is a fine-tuned version of [openai/small](https://huggingface.co/openai/small) on the child-50k dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.0241
|
| 20 |
+
- Wer: 2.2556
|
| 21 |
+
- Cer: 0.9159
|
| 22 |
+
|
| 23 |
+
## Model description
|
| 24 |
+
|
| 25 |
+
More information needed
|
| 26 |
+
|
| 27 |
+
## Intended uses & limitations
|
| 28 |
+
|
| 29 |
+
More information needed
|
| 30 |
+
|
| 31 |
+
## Training and evaluation data
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Training procedure
|
| 36 |
+
|
| 37 |
+
### Training hyperparameters
|
| 38 |
+
|
| 39 |
+
The following hyperparameters were used during training:
|
| 40 |
+
- learning_rate: 1.25e-05
|
| 41 |
+
- train_batch_size: 16
|
| 42 |
+
- eval_batch_size: 8
|
| 43 |
+
- seed: 42
|
| 44 |
+
- gradient_accumulation_steps: 2
|
| 45 |
+
- total_train_batch_size: 32
|
| 46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
+
- lr_scheduler_type: linear
|
| 48 |
+
- lr_scheduler_warmup_steps: 500
|
| 49 |
+
- training_steps: 5000
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 54 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 55 |
+
| 0.1058 | 0.36 | 500 | 0.0708 | 8.0241 | 3.2266 |
|
| 56 |
+
| 0.0577 | 0.71 | 1000 | 0.0446 | 5.2591 | 2.1891 |
|
| 57 |
+
| 0.0296 | 1.07 | 1500 | 0.0309 | 3.2662 | 1.3469 |
|
| 58 |
+
| 0.0301 | 1.42 | 2000 | 0.0279 | 2.7286 | 1.1516 |
|
| 59 |
+
| 0.0299 | 1.78 | 2500 | 0.0245 | 2.4578 | 0.9920 |
|
| 60 |
+
| 0.0219 | 2.13 | 3000 | 0.0262 | 2.6882 | 1.2290 |
|
| 61 |
+
| 0.0148 | 2.49 | 3500 | 0.0219 | 2.0899 | 0.9245 |
|
| 62 |
+
| 0.0141 | 2.84 | 4000 | 0.0281 | 2.8782 | 1.1615 |
|
| 63 |
+
| 0.0097 | 3.2 | 4500 | 0.0210 | 2.0616 | 0.8373 |
|
| 64 |
+
| 0.0109 | 3.55 | 5000 | 0.0241 | 2.2556 | 0.9159 |
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
### Framework versions
|
| 68 |
+
|
| 69 |
+
- Transformers 4.34.0
|
| 70 |
+
- Pytorch 2.1.0+cu121
|
| 71 |
+
- Datasets 2.14.5
|
| 72 |
+
- Tokenizers 0.14.1
|