Commit
•
4ed44bb
1
Parent(s):
da9f196
update model card
Browse files
README.md
CHANGED
@@ -9,14 +9,10 @@ tags:
|
|
9 |
- generated_from_trainer
|
10 |
datasets:
|
11 |
- mozilla-foundation/common_voice_11_0
|
12 |
-
- mozilla-foundation/common_voice_11_0
|
13 |
-
- mozilla-foundation/common_voice_11_0
|
14 |
- babelbox/babelbox_voice
|
15 |
- NbAiLab/NST
|
16 |
- NbAiLab/NPSC
|
17 |
- google/fleurs
|
18 |
-
- google/fleurs
|
19 |
-
- google/fleurs
|
20 |
metrics:
|
21 |
- wer
|
22 |
model-index:
|
@@ -33,47 +29,34 @@ model-index:
|
|
33 |
metrics:
|
34 |
- name: Wer
|
35 |
type: wer
|
36 |
-
value: 11.
|
37 |
-
- task:
|
38 |
-
name: Automatic Speech Recognition
|
39 |
-
type: automatic-speech-recognition
|
40 |
-
dataset:
|
41 |
-
name: babelbox/babelbox_voice
|
42 |
-
type: babelbox/babelbox_voice
|
43 |
-
metrics:
|
44 |
-
- name: Wer
|
45 |
-
type: wer
|
46 |
-
value: 11.307923879152778
|
47 |
- task:
|
48 |
name: Automatic Speech Recognition
|
49 |
type: automatic-speech-recognition
|
50 |
dataset:
|
51 |
-
name:
|
52 |
-
type:
|
53 |
-
|
54 |
-
|
55 |
-
type: wer
|
56 |
-
value: 11.307923879152778
|
57 |
-
- task:
|
58 |
-
name: Automatic Speech Recognition
|
59 |
-
type: automatic-speech-recognition
|
60 |
-
dataset:
|
61 |
-
name: NbAiLab/NPSC
|
62 |
-
type: NbAiLab/NPSC
|
63 |
metrics:
|
64 |
- name: Wer
|
65 |
type: wer
|
66 |
-
value:
|
67 |
- task:
|
68 |
name: Automatic Speech Recognition
|
69 |
type: automatic-speech-recognition
|
70 |
dataset:
|
71 |
-
name:
|
72 |
-
type:
|
|
|
|
|
73 |
metrics:
|
74 |
- name: Wer
|
75 |
type: wer
|
76 |
-
value:
|
|
|
|
|
|
|
77 |
---
|
78 |
|
79 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -81,8 +64,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
81 |
|
82 |
# Whisper Medium Nordic
|
83 |
|
84 |
-
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the mozilla-foundation/common_voice_11_0
|
85 |
-
|
|
|
86 |
- Loss: 0.2129
|
87 |
- Wer: 11.3079
|
88 |
|
@@ -100,6 +84,9 @@ More information needed
|
|
100 |
|
101 |
## Training procedure
|
102 |
|
|
|
|
|
|
|
103 |
### Training hyperparameters
|
104 |
|
105 |
The following hyperparameters were used during training:
|
|
|
9 |
- generated_from_trainer
|
10 |
datasets:
|
11 |
- mozilla-foundation/common_voice_11_0
|
|
|
|
|
12 |
- babelbox/babelbox_voice
|
13 |
- NbAiLab/NST
|
14 |
- NbAiLab/NPSC
|
15 |
- google/fleurs
|
|
|
|
|
16 |
metrics:
|
17 |
- wer
|
18 |
model-index:
|
|
|
29 |
metrics:
|
30 |
- name: Wer
|
31 |
type: wer
|
32 |
+
value: 11.31
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
- task:
|
34 |
name: Automatic Speech Recognition
|
35 |
type: automatic-speech-recognition
|
36 |
dataset:
|
37 |
+
name: mozilla-foundation/common_voice_11_0
|
38 |
+
type: mozilla-foundation/common_voice_11_0
|
39 |
+
config: da
|
40 |
+
split: test
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
metrics:
|
42 |
- name: Wer
|
43 |
type: wer
|
44 |
+
value: 14.86
|
45 |
- task:
|
46 |
name: Automatic Speech Recognition
|
47 |
type: automatic-speech-recognition
|
48 |
dataset:
|
49 |
+
name: mozilla-foundation/common_voice_11_0
|
50 |
+
type: mozilla-foundation/common_voice_11_0
|
51 |
+
config: nn-NO
|
52 |
+
split: test
|
53 |
metrics:
|
54 |
- name: Wer
|
55 |
type: wer
|
56 |
+
value: 37.02
|
57 |
+
|
58 |
+
|
59 |
+
|
60 |
---
|
61 |
|
62 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
64 |
|
65 |
# Whisper Medium Nordic
|
66 |
|
67 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the [mozilla-foundation/common_voice_11_0](https://huggingface.co/datasets/mozilla-foundation/common_voice_11_0) (sv-SE, da, nn-NO), the [babelbox/babelbox_voice](https://huggingface.co/datasets/babelbox/babelbox_voice) (Swedish radio), the [NbAiLab/NST](https://huggingface.co/datasets/NbAiLab/NST) (Norwegian radio), the [NbAiLab/NPSC](https://huggingface.co/datasets/NbAiLab/NPSC) (Norwegian parliament) and the [google/fleurs](https://huggingface.co/datasets/google/fleurs) (sv_se, da_dk, nb_no) datasets. The goal is to leverage transfer learning across Nordic languages, which have strong similarities.
|
68 |
+
|
69 |
+
It achieves the following results on the common voice Swedish test set:
|
70 |
- Loss: 0.2129
|
71 |
- Wer: 11.3079
|
72 |
|
|
|
84 |
|
85 |
## Training procedure
|
86 |
|
87 |
+
Please note that a bug during training prevented us from evaluating WER correctly.
|
88 |
+
Validation loss suggests we started overfitting after 5000/6000 steps.
|
89 |
+
|
90 |
### Training hyperparameters
|
91 |
|
92 |
The following hyperparameters were used during training:
|