agnesluhtaru
commited on
Commit
•
19385f6
1
Parent(s):
13067d9
update model card README.md
Browse files
README.md
CHANGED
@@ -1,30 +1,23 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
tags:
|
4 |
-
-
|
5 |
metrics:
|
6 |
- wer
|
7 |
model-index:
|
8 |
-
- name:
|
9 |
-
results:
|
10 |
-
- task:
|
11 |
-
type: automatic-speech-recognition
|
12 |
-
name: Automatic Speech Recognition
|
13 |
-
dataset:
|
14 |
-
name: mozilla-foundation/common_voice_11_0
|
15 |
-
type: mozilla-foundation/common_voice_11_0
|
16 |
-
config: et
|
17 |
-
split: test
|
18 |
-
metrics:
|
19 |
-
- type: wer
|
20 |
-
value: 20.56
|
21 |
-
name: WER
|
22 |
---
|
23 |
|
24 |
-
|
|
|
25 |
|
26 |
-
|
27 |
-
|
|
|
|
|
|
|
|
|
28 |
|
29 |
## Model description
|
30 |
|
@@ -36,7 +29,7 @@ More information needed
|
|
36 |
|
37 |
## Training and evaluation data
|
38 |
|
39 |
-
|
40 |
|
41 |
## Training procedure
|
42 |
|
@@ -50,12 +43,28 @@ The following hyperparameters were used during training:
|
|
50 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
51 |
- lr_scheduler_type: linear
|
52 |
- lr_scheduler_warmup_steps: 500
|
53 |
-
- training_steps:
|
54 |
- mixed_precision_training: Native AMP
|
55 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
### Framework versions
|
57 |
|
58 |
- Transformers 4.26.0.dev0
|
59 |
- Pytorch 1.12.1+rocm5.1.1
|
60 |
- Datasets 2.7.1.dev0
|
61 |
-
- Tokenizers 0.13.2
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
tags:
|
4 |
+
- generated_from_trainer
|
5 |
metrics:
|
6 |
- wer
|
7 |
model-index:
|
8 |
+
- name: whisper-medium-et-ERR2020
|
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 |
+
# whisper-medium-et-ERR2020
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the None dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 0.3672
|
20 |
+
- Wer: 15.9119
|
21 |
|
22 |
## Model description
|
23 |
|
|
|
29 |
|
30 |
## Training and evaluation data
|
31 |
|
32 |
+
More information needed
|
33 |
|
34 |
## Training procedure
|
35 |
|
|
|
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 |
+
- training_steps: 10000
|
47 |
- mixed_precision_training: Native AMP
|
48 |
|
49 |
+
### Training results
|
50 |
+
|
51 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
52 |
+
|:-------------:|:-----:|:-----:|:---------------:|:-------:|
|
53 |
+
| 0.1828 | 0.1 | 1000 | 0.3547 | 20.8829 |
|
54 |
+
| 0.09 | 0.2 | 2000 | 0.3476 | 19.0096 |
|
55 |
+
| 0.083 | 0.3 | 3000 | 0.3386 | 18.1304 |
|
56 |
+
| 0.0765 | 0.4 | 4000 | 0.3365 | 17.2591 |
|
57 |
+
| 0.0592 | 0.5 | 5000 | 0.3534 | 19.0213 |
|
58 |
+
| 0.0672 | 0.6 | 6000 | 0.3622 | 18.4263 |
|
59 |
+
| 0.0629 | 0.7 | 7000 | 0.3487 | 15.9839 |
|
60 |
+
| 0.0546 | 1.03 | 8000 | 0.3677 | 16.1021 |
|
61 |
+
| 0.0459 | 1.13 | 9000 | 0.3704 | 17.9073 |
|
62 |
+
| 0.0425 | 1.23 | 10000 | 0.3672 | 15.9119 |
|
63 |
+
|
64 |
+
|
65 |
### Framework versions
|
66 |
|
67 |
- Transformers 4.26.0.dev0
|
68 |
- Pytorch 1.12.1+rocm5.1.1
|
69 |
- Datasets 2.7.1.dev0
|
70 |
+
- Tokenizers 0.13.2
|