Create README.md
Browse files
README.md
CHANGED
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- el
|
4 |
+
license: apache-2.0
|
5 |
+
tags:
|
6 |
+
- whisper-event
|
7 |
+
datasets:
|
8 |
+
- mozilla-foundation/common_voice_11_0
|
9 |
+
metrics:
|
10 |
+
- wer
|
11 |
+
model-index:
|
12 |
+
- name: whisper-small-el
|
13 |
+
results:
|
14 |
+
- task:
|
15 |
+
name: Automatic Speech Recognition
|
16 |
+
type: automatic-speech-recognition
|
17 |
+
dataset:
|
18 |
+
name: mozilla-foundation/common_voice_11_0 el
|
19 |
+
type: mozilla-foundation/common_voice_11_0
|
20 |
+
config: el
|
21 |
+
split: test
|
22 |
+
args: el
|
23 |
+
metrics:
|
24 |
+
- name: Wer
|
25 |
+
type: wer
|
26 |
+
value: 25.696508172362552
|
27 |
+
---
|
28 |
+
|
29 |
+
# Whisper Small - Greek (el)
|
30 |
+
|
31 |
+
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the mozilla-foundation/common_voice_11_0 el dataset
|
32 |
+
for transcription in Greek.
|
33 |
+
It achieves the following results on the evaluation set:
|
34 |
+
|
35 |
+
- Wer: 20.2080
|
36 |
+
|
37 |
+
|
38 |
+
### Training results
|
39 |
+
|
40 |
+
Upon completion of training the best model was reloaded and tested with the following results extracted from the stdout log:
|
41 |
+
```
|
42 |
+
Loading best model from ./whisper-small-el/checkpoint-5000 (score: 20.208023774145616).
|
43 |
+
|
44 |
+
{'train_runtime': 73232.697,
|
45 |
+
'train_samples_per_second': 4.37,
|
46 |
+
'train_steps_per_second': 0.068,
|
47 |
+
'train_loss': 0.06146362095708027,
|
48 |
+
'epoch': 94.34}
|
49 |
+
|
50 |
+
TrainOutput(global_step=5000,
|
51 |
+
training_loss=0.06146362095708027,
|
52 |
+
metrics={'train_runtime': 73232.697,
|
53 |
+
'train_samples_per_second': 4.37,
|
54 |
+
'train_steps_per_second': 0.068,
|
55 |
+
'train_loss': 0.06146362095708027,
|
56 |
+
'epoch': 94.34})
|
57 |
+
```
|
58 |
+
|
59 |
+
### Framework versions
|
60 |
+
|
61 |
+
- Transformers 4.26.0.dev0
|
62 |
+
- Pytorch 1.13.0
|
63 |
+
- Datasets 2.7.1.dev0
|
64 |
+
- Tokenizers 0.12.1
|