Commit
•
09dedd1
1
Parent(s):
50b2b69
Update README.md
Browse files
README.md
CHANGED
|
@@ -11,14 +11,21 @@ datasets:
|
|
| 11 |
model-index:
|
| 12 |
- name: whisper-small_child50K_timestretch_stepLR
|
| 13 |
results: []
|
|
|
|
|
|
|
|
|
|
| 14 |
---
|
| 15 |
|
| 16 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 17 |
should probably proofread and complete it, then remove this comment. -->
|
| 18 |
|
| 19 |
-
# whisper-
|
| 20 |
|
| 21 |
-
This model is a fine-tuned version of [openai/whisper-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
@@ -46,9 +53,25 @@ The following hyperparameters were used during training:
|
|
| 46 |
- lr_scheduler_warmup_steps: 500
|
| 47 |
- training_steps: 5000
|
| 48 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
### Framework versions
|
| 50 |
|
| 51 |
- Transformers 4.34.1
|
| 52 |
- Pytorch 1.12.1
|
| 53 |
- Datasets 2.14.5
|
| 54 |
-
- Tokenizers 0.14.1
|
|
|
|
| 11 |
model-index:
|
| 12 |
- name: whisper-small_child50K_timestretch_stepLR
|
| 13 |
results: []
|
| 14 |
+
metrics:
|
| 15 |
+
- wer
|
| 16 |
+
- cer
|
| 17 |
---
|
| 18 |
|
| 19 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 20 |
should probably proofread and complete it, then remove this comment. -->
|
| 21 |
|
| 22 |
+
# whisper-small_child50K_timeStretch_stepLR
|
| 23 |
|
| 24 |
+
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the child-50k dataset.
|
| 25 |
+
It achieves the following results on the evaluation set:
|
| 26 |
+
- Loss: 0.0201
|
| 27 |
+
- Wer: 2.1061
|
| 28 |
+
- Cer: 0.8189
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
|
|
| 53 |
- lr_scheduler_warmup_steps: 500
|
| 54 |
- training_steps: 5000
|
| 55 |
|
| 56 |
+
### Training results
|
| 57 |
+
|
| 58 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 59 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 60 |
+
| 0.1004 | 0.18 | 500 | 0.0796 | 9.5440 | 4.0921 |
|
| 61 |
+
| 0.0846 | 0.36 | 1000 | 0.0453 | 5.3319 | 2.3843 |
|
| 62 |
+
| 0.0729 | 0.53 | 1500 | 0.0355 | 4.2849 | 1.7311 |
|
| 63 |
+
| 0.0486 | 0.71 | 2000 | 0.0284 | 2.8701 | 1.2241 |
|
| 64 |
+
| 0.045 | 0.89 | 2500 | 0.0261 | 3.6220 | 2.6581 |
|
| 65 |
+
| 0.0206 | 1.07 | 3000 | 0.0207 | 2.0616 | 0.8263 |
|
| 66 |
+
| 0.0264 | 1.24 | 3500 | 0.0219 | 2.0091 | 0.8275 |
|
| 67 |
+
| 0.0304 | 1.42 | 4000 | 0.0205 | 1.7827 | 0.7207 |
|
| 68 |
+
| 0.0224 | 1.6 | 4500 | 0.0233 | 2.3527 | 0.9527 |
|
| 69 |
+
| 0.0215 | 1.78 | 5000 | 0.0201 | 2.1061 | 0.8189 |
|
| 70 |
+
|
| 71 |
+
|
| 72 |
### Framework versions
|
| 73 |
|
| 74 |
- Transformers 4.34.1
|
| 75 |
- Pytorch 1.12.1
|
| 76 |
- Datasets 2.14.5
|
| 77 |
+
- Tokenizers 0.14.1
|