marinone94
commited on
Commit
•
628520f
1
Parent(s):
bb89fa3
update model card README.md
Browse files
README.md
CHANGED
@@ -1,47 +1,12 @@
|
|
1 |
---
|
2 |
-
language:
|
3 |
-
- sv-SE
|
4 |
license: cc0-1.0
|
5 |
tags:
|
6 |
-
- automatic-speech-recognition
|
7 |
-
- mozilla-foundation/common_voice_7_0
|
8 |
-
- sv
|
9 |
- generated_from_trainer
|
10 |
-
- robust-speech-event
|
11 |
-
- model_for_talk
|
12 |
datasets:
|
13 |
-
-
|
14 |
model-index:
|
15 |
-
- name:
|
16 |
-
results:
|
17 |
-
- task:
|
18 |
-
name: Automatic Speech Recognition
|
19 |
-
type: automatic-speech-recognition
|
20 |
-
dataset:
|
21 |
-
name: Common Voice 7
|
22 |
-
type: mozilla-foundation/common_voice_7_0
|
23 |
-
args: sv-SE
|
24 |
-
metrics:
|
25 |
-
- name: Test WER
|
26 |
-
type: wer
|
27 |
-
value: 18.89
|
28 |
-
- name: Test CER
|
29 |
-
type: cer
|
30 |
-
value: 6.63
|
31 |
-
- task:
|
32 |
-
name: Automatic Speech Recognition
|
33 |
-
type: automatic-speech-recognition
|
34 |
-
dataset:
|
35 |
-
name: Robust Speech Event - Dev Data
|
36 |
-
type: speech-recognition-community-v2/dev_data
|
37 |
-
args: sv
|
38 |
-
metrics:
|
39 |
-
- name: Test WER
|
40 |
-
type: wer
|
41 |
-
value: 30.65
|
42 |
-
- name: Test CER
|
43 |
-
type: cer
|
44 |
-
value: 13.56
|
45 |
---
|
46 |
|
47 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -49,10 +14,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
49 |
|
50 |
#
|
51 |
|
52 |
-
This model is a fine-tuned version of [KBLab/wav2vec2-large-voxrex](https://huggingface.co/KBLab/wav2vec2-large-voxrex) on the
|
53 |
It achieves the following results on the evaluation set:
|
54 |
-
- Loss:
|
55 |
-
- Wer: 0
|
56 |
|
57 |
## Model description
|
58 |
|
@@ -71,61 +36,64 @@ More information needed
|
|
71 |
### Training hyperparameters
|
72 |
|
73 |
The following hyperparameters were used during training:
|
74 |
-
- learning_rate:
|
75 |
-
- train_batch_size:
|
76 |
-
- eval_batch_size:
|
77 |
- seed: 42
|
78 |
- gradient_accumulation_steps: 4
|
79 |
-
- total_train_batch_size:
|
80 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
81 |
- lr_scheduler_type: linear
|
82 |
-
-
|
83 |
-
- num_epochs:
|
84 |
- mixed_precision_training: Native AMP
|
85 |
|
86 |
### Training results
|
87 |
|
88 |
-
| Training Loss | Epoch | Step
|
89 |
-
|
90 |
-
| 3.
|
91 |
-
|
|
92 |
-
|
|
93 |
-
|
|
94 |
-
|
|
95 |
-
|
|
96 |
-
|
|
97 |
-
|
|
98 |
-
|
|
99 |
-
|
|
100 |
-
|
|
101 |
-
|
|
102 |
-
|
|
103 |
-
|
|
104 |
-
|
|
105 |
-
|
|
106 |
-
|
|
107 |
-
|
|
108 |
-
|
|
109 |
-
|
|
110 |
-
|
|
111 |
-
|
|
112 |
-
|
|
113 |
-
|
|
114 |
-
|
|
115 |
-
|
|
116 |
-
|
|
117 |
-
|
|
118 |
-
|
|
119 |
-
|
|
120 |
-
|
|
121 |
-
|
|
122 |
-
|
|
123 |
-
|
|
|
|
|
|
|
|
124 |
|
125 |
|
126 |
### Framework versions
|
127 |
|
128 |
-
- Transformers 4.
|
129 |
-
- Pytorch 1.10.
|
130 |
-
- Datasets 1.
|
131 |
- Tokenizers 0.11.0
|
|
|
1 |
---
|
|
|
|
|
2 |
license: cc0-1.0
|
3 |
tags:
|
|
|
|
|
|
|
4 |
- generated_from_trainer
|
|
|
|
|
5 |
datasets:
|
6 |
+
- nst_sv
|
7 |
model-index:
|
8 |
+
- name: ''
|
9 |
+
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
---
|
11 |
|
12 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
14 |
|
15 |
#
|
16 |
|
17 |
+
This model is a fine-tuned version of [KBLab/wav2vec2-large-voxrex](https://huggingface.co/KBLab/wav2vec2-large-voxrex) on the nst_sv dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: inf
|
20 |
+
- Wer: 1.0
|
21 |
|
22 |
## Model description
|
23 |
|
|
|
36 |
### Training hyperparameters
|
37 |
|
38 |
The following hyperparameters were used during training:
|
39 |
+
- learning_rate: 0.00075
|
40 |
+
- train_batch_size: 32
|
41 |
+
- eval_batch_size: 32
|
42 |
- seed: 42
|
43 |
- gradient_accumulation_steps: 4
|
44 |
+
- total_train_batch_size: 128
|
45 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
- lr_scheduler_type: linear
|
47 |
+
- lr_scheduler_warmup_ratio: 0.02
|
48 |
+
- num_epochs: 2.0
|
49 |
- mixed_precision_training: Native AMP
|
50 |
|
51 |
### Training results
|
52 |
|
53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
54 |
+
|:-------------:|:-----:|:----:|:---------------:|:---:|
|
55 |
+
| 3.4039 | 0.05 | 100 | inf | 1.0 |
|
56 |
+
| 3.4396 | 0.11 | 200 | inf | 1.0 |
|
57 |
+
| 3.483 | 0.16 | 300 | inf | 1.0 |
|
58 |
+
| 3.5014 | 0.21 | 400 | inf | 1.0 |
|
59 |
+
| 3.331 | 0.27 | 500 | inf | 1.0 |
|
60 |
+
| 3.4809 | 0.32 | 600 | inf | 1.0 |
|
61 |
+
| 3.4678 | 0.37 | 700 | inf | 1.0 |
|
62 |
+
| 3.4596 | 0.43 | 800 | inf | 1.0 |
|
63 |
+
| 3.4644 | 0.48 | 900 | inf | 1.0 |
|
64 |
+
| 3.4671 | 0.53 | 1000 | inf | 1.0 |
|
65 |
+
| 3.6005 | 0.59 | 1100 | inf | 1.0 |
|
66 |
+
| 3.9182 | 0.64 | 1200 | inf | 1.0 |
|
67 |
+
| 3.6466 | 0.69 | 1300 | inf | 1.0 |
|
68 |
+
| 3.6932 | 0.75 | 1400 | inf | 1.0 |
|
69 |
+
| 3.7939 | 0.8 | 1500 | inf | 1.0 |
|
70 |
+
| 3.9284 | 0.85 | 1600 | inf | 1.0 |
|
71 |
+
| 3.7859 | 0.91 | 1700 | inf | 1.0 |
|
72 |
+
| 3.9363 | 0.96 | 1800 | inf | 1.0 |
|
73 |
+
| 3.7573 | 1.01 | 1900 | inf | 1.0 |
|
74 |
+
| 3.7553 | 1.07 | 2000 | inf | 1.0 |
|
75 |
+
| 3.7606 | 1.12 | 2100 | inf | 1.0 |
|
76 |
+
| 3.7514 | 1.17 | 2200 | inf | 1.0 |
|
77 |
+
| 3.7472 | 1.23 | 2300 | inf | 1.0 |
|
78 |
+
| 3.7478 | 1.28 | 2400 | inf | 1.0 |
|
79 |
+
| 3.7496 | 1.33 | 2500 | inf | 1.0 |
|
80 |
+
| 3.7513 | 1.39 | 2600 | inf | 1.0 |
|
81 |
+
| 3.7497 | 1.44 | 2700 | inf | 1.0 |
|
82 |
+
| 3.7539 | 1.49 | 2800 | inf | 1.0 |
|
83 |
+
| 3.7581 | 1.55 | 2900 | inf | 1.0 |
|
84 |
+
| 3.7572 | 1.6 | 3000 | inf | 1.0 |
|
85 |
+
| 3.7589 | 1.66 | 3100 | inf | 1.0 |
|
86 |
+
| 3.7592 | 1.71 | 3200 | inf | 1.0 |
|
87 |
+
| 3.7531 | 1.76 | 3300 | inf | 1.0 |
|
88 |
+
| 3.7567 | 1.82 | 3400 | inf | 1.0 |
|
89 |
+
| 3.7613 | 1.87 | 3500 | inf | 1.0 |
|
90 |
+
| 3.7516 | 1.92 | 3600 | inf | 1.0 |
|
91 |
+
| 3.7581 | 1.98 | 3700 | inf | 1.0 |
|
92 |
|
93 |
|
94 |
### Framework versions
|
95 |
|
96 |
+
- Transformers 4.17.0.dev0
|
97 |
+
- Pytorch 1.10.2+cu102
|
98 |
+
- Datasets 1.18.3
|
99 |
- Tokenizers 0.11.0
|