Commit
•
29a8690
1
Parent(s):
15ff319
update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
metrics:
|
| 6 |
+
- wer
|
| 7 |
+
model-index:
|
| 8 |
+
- name: KalemaTech-Arabic-STT-ASR-based-on-Whisper-Small
|
| 9 |
+
results: []
|
| 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 |
+
# KalemaTech-Arabic-STT-ASR-based-on-Whisper-Small
|
| 16 |
+
|
| 17 |
+
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on an unknown dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.5362
|
| 20 |
+
- Wer: 58.5848
|
| 21 |
+
|
| 22 |
+
## Model description
|
| 23 |
+
|
| 24 |
+
More information needed
|
| 25 |
+
|
| 26 |
+
## Intended uses & limitations
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Training and evaluation data
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Training procedure
|
| 35 |
+
|
| 36 |
+
### Training hyperparameters
|
| 37 |
+
|
| 38 |
+
The following hyperparameters were used during training:
|
| 39 |
+
- learning_rate: 1e-05
|
| 40 |
+
- train_batch_size: 64
|
| 41 |
+
- eval_batch_size: 8
|
| 42 |
+
- seed: 42
|
| 43 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 44 |
+
- lr_scheduler_type: linear
|
| 45 |
+
- lr_scheduler_warmup_steps: 500
|
| 46 |
+
- num_epochs: 25
|
| 47 |
+
- mixed_precision_training: Native AMP
|
| 48 |
+
|
| 49 |
+
### Training results
|
| 50 |
+
|
| 51 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 52 |
+
|:-------------:|:-----:|:-----:|:---------------:|:-------:|
|
| 53 |
+
| 0.2728 | 1.01 | 1000 | 0.3063 | 60.4733 |
|
| 54 |
+
| 0.1442 | 2.01 | 2000 | 0.2878 | 55.6935 |
|
| 55 |
+
| 0.0648 | 3.02 | 3000 | 0.3009 | 59.2568 |
|
| 56 |
+
| 0.0318 | 4.03 | 4000 | 0.3278 | 59.2993 |
|
| 57 |
+
| 0.0148 | 5.04 | 5000 | 0.3539 | 61.0364 |
|
| 58 |
+
| 0.0088 | 6.04 | 6000 | 0.3714 | 56.9154 |
|
| 59 |
+
| 0.0061 | 7.05 | 7000 | 0.3920 | 57.5515 |
|
| 60 |
+
| 0.0041 | 8.06 | 8000 | 0.4149 | 61.6328 |
|
| 61 |
+
| 0.0033 | 9.06 | 9000 | 0.4217 | 58.0310 |
|
| 62 |
+
| 0.0033 | 10.07 | 10000 | 0.4376 | 59.9594 |
|
| 63 |
+
| 0.0021 | 11.08 | 11000 | 0.4485 | 56.7812 |
|
| 64 |
+
| 0.0015 | 12.08 | 12000 | 0.4577 | 57.6936 |
|
| 65 |
+
| 0.0013 | 13.09 | 13000 | 0.4671 | 60.6606 |
|
| 66 |
+
| 0.0011 | 14.1 | 14000 | 0.4686 | 59.8159 |
|
| 67 |
+
| 0.0008 | 15.11 | 15000 | 0.4856 | 60.7111 |
|
| 68 |
+
| 0.0011 | 16.11 | 16000 | 0.4851 | 59.5198 |
|
| 69 |
+
| 0.0005 | 17.12 | 17000 | 0.4936 | 59.2608 |
|
| 70 |
+
| 0.0004 | 18.13 | 18000 | 0.4995 | 57.9619 |
|
| 71 |
+
| 0.0003 | 19.13 | 19000 | 0.5085 | 58.3630 |
|
| 72 |
+
| 0.0002 | 20.14 | 20000 | 0.5155 | 58.0987 |
|
| 73 |
+
| 0.0001 | 21.15 | 21000 | 0.5251 | 58.8504 |
|
| 74 |
+
| 0.0001 | 22.16 | 22000 | 0.5268 | 58.4228 |
|
| 75 |
+
| 0.0001 | 23.16 | 23000 | 0.5317 | 59.0881 |
|
| 76 |
+
| 0.0001 | 24.17 | 24000 | 0.5362 | 58.5848 |
|
| 77 |
+
|
| 78 |
+
|
| 79 |
+
### Framework versions
|
| 80 |
+
|
| 81 |
+
- Transformers 4.25.1
|
| 82 |
+
- Pytorch 1.13.1+cu117
|
| 83 |
+
- Datasets 2.8.0
|
| 84 |
+
- Tokenizers 0.13.2
|