Hugging Face's logo

arampacha
/
wav2vec2-xls-r-1b-uk

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Ukrainian
wav2vec2
Generated from Trainer
hf-asr-leaderboard
mozilla-foundation/common_voice_8_0
robust-speech-event
Eval Results
Inference Endpoints
Model card Files Metrics Community
2
arampacha commited on
Commit
4d73755
•
1 Parent(s): 9ff1856

upd readme

Browse files
Files changed (3) hide show
  1. README.md +53 -30
  2. log_speech-recognition-community-v2_dev_data_uk_validation_predictions.txt +0 -0
  3. log_speech-recognition-community-v2_dev_data_uk_validation_targets.txt +0 -0
README.md CHANGED
@@ -1,12 +1,31 @@
1
  ---
 
 
2
  license: apache-2.0
3
  tags:
4
  - automatic-speech-recognition
5
- - /workspace/data/uk/noizy_student_1/
6
  - generated_from_trainer
 
 
 
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
@@ -14,11 +33,11 @@ should probably proofread and complete it, then remove this comment. -->
14
 
15
  #
16
 
17
- This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on the /WORKSPACE/DATA/UK/NOIZY_STUDENT_1/ - NA dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.1285
20
- - Wer: 0.1821
21
- - Cer: 0.0342
22
 
23
  ## Model description
24
 
@@ -46,33 +65,37 @@ The following hyperparameters were used during training:
46
  - optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
47
  - lr_scheduler_type: cosine
48
  - lr_scheduler_warmup_ratio: 0.1
49
- - training_steps: 10000
50
  - mixed_precision_training: Native AMP
51
 
52
  ### Training results
53
 
54
- | Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
55
- |:-------------:|:-----:|:-----:|:------:|:---------------:|:------:|
56
- | 1.2323 | 3.22 | 500 | 0.0797 | 0.2816 | 0.4133 |
57
- | 0.9826 | 6.45 | 1000 | 0.0514 | 0.1970 | 0.2688 |
58
- | 0.8628 | 9.67 | 1500 | 0.0474 | 0.1649 | 0.2485 |
59
- | 0.8348 | 12.9 | 2000 | 0.0467 | 0.1605 | 0.2460 |
60
- | 0.8186 | 16.13 | 2500 | 0.0469 | 0.1608 | 0.2469 |
61
- | 0.8011 | 19.35 | 3000 | 0.1620 | 0.2412 | 0.0468 |
62
- | 0.807 | 22.58 | 3500 | 0.1737 | 0.2524 | 0.0498 |
63
- | 0.7758 | 25.8 | 4000 | 0.1709 | 0.2536 | 0.0498 |
64
- | 0.7923 | 29.03 | 4500 | 0.1645 | 0.2436 | 0.0474 |
65
- | 0.7717 | 32.26 | 5000 | 0.1811 | 0.2636 | 0.0524 |
66
- | 0.7447 | 35.48 | 5500 | 0.1635 | 0.2405 | 0.0468 |
67
- | 0.7267 | 38.71 | 6000 | 0.1578 | 0.2354 | 0.0462 |
68
- | 0.7046 | 41.93 | 6500 | 0.1555 | 0.2296 | 0.0444 |
69
- | 0.6896 | 45.16 | 7000 | 0.1548 | 0.2272 | 0.0439 |
70
- | 0.6575 | 48.38 | 7500 | 0.1432 | 0.2096 | 0.0399 |
71
- | 0.6264 | 51.61 | 8000 | 0.1466 | 0.2056 | 0.0398 |
72
- | 0.589 | 54.83 | 8500 | 0.1351 | 0.1943 | 0.0371 |
73
- | 0.573 | 58.06 | 9000 | 0.1387 | 0.1934 | 0.0365 |
74
- | 0.5537 | 61.29 | 9500 | 0.1328 | 0.1883 | 0.0353 |
75
- | 0.544 | 64.51 | 10000 | 0.1285 | 0.1821 | 0.0342 |
 
 
 
 
76
 
77
 
78
  ### Framework versions
 
1
  ---
2
+ language:
3
+ - uk
4
  license: apache-2.0
5
  tags:
6
  - automatic-speech-recognition
7
+ - mozilla-foundation/common_voice_8_0
8
  - generated_from_trainer
9
+ - robust-speech-event
10
+ datasets:
11
+ - common_voice
12
  model-index:
13
+ - name: wav2vec2-xls-r-1b-hy
14
+ results:
15
+ - task:
16
+ type: automatic-speech-recognition
17
+ name: Speech Recognition
18
+ dataset:
19
+ type: mozilla-foundation/common_voice_8_0
20
+ name: Common Voice uk
21
+ args: uk
22
+ metrics:
23
+ - type: wer
24
+ value: 10.406342913776016
25
+ name: WER LM
26
+ - type: cer
27
+ value: 2.0387492208601702
28
+ name: CER LM
29
  ---
30
 
31
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  #
35
 
36
+ This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on the /WORKSPACE/DATA/UK/COMPOSED_DATASET/ - NA dataset.
37
  It achieves the following results on the evaluation set:
38
+ - Loss: 0.1092
39
+ - Wer: 0.1752
40
+ - Cer: 0.0323
41
 
42
  ## Model description
43
 
 
65
  - optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
66
  - lr_scheduler_type: cosine
67
  - lr_scheduler_warmup_ratio: 0.1
68
+ - training_steps: 12000
69
  - mixed_precision_training: Native AMP
70
 
71
  ### Training results
72
 
73
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
74
+ |:-------------:|:-----:|:-----:|:---------------:|:------:|:------:|
75
+ | 1.7005 | 1.61 | 500 | 0.4082 | 0.5584 | 0.1164 |
76
+ | 1.1555 | 3.22 | 1000 | 0.2020 | 0.2953 | 0.0557 |
77
+ | 1.0927 | 4.82 | 1500 | 0.1708 | 0.2584 | 0.0480 |
78
+ | 1.0707 | 6.43 | 2000 | 0.1563 | 0.2405 | 0.0450 |
79
+ | 1.0728 | 8.04 | 2500 | 0.1620 | 0.2442 | 0.0463 |
80
+ | 1.0268 | 9.65 | 3000 | 0.1588 | 0.2378 | 0.0458 |
81
+ | 1.0328 | 11.25 | 3500 | 0.1466 | 0.2352 | 0.0442 |
82
+ | 1.0249 | 12.86 | 4000 | 0.1552 | 0.2341 | 0.0449 |
83
+ | 1.016 | 14.47 | 4500 | 0.1602 | 0.2435 | 0.0473 |
84
+ | 1.0164 | 16.08 | 5000 | 0.1491 | 0.2337 | 0.0444 |
85
+ | 0.9935 | 17.68 | 5500 | 0.1539 | 0.2373 | 0.0458 |
86
+ | 0.9626 | 19.29 | 6000 | 0.1458 | 0.2305 | 0.0434 |
87
+ | 0.9505 | 20.9 | 6500 | 0.1368 | 0.2157 | 0.0407 |
88
+ | 0.9389 | 22.51 | 7000 | 0.1437 | 0.2231 | 0.0426 |
89
+ | 0.9129 | 24.12 | 7500 | 0.1313 | 0.2076 | 0.0394 |
90
+ | 0.9118 | 25.72 | 8000 | 0.1292 | 0.2040 | 0.0384 |
91
+ | 0.8848 | 27.33 | 8500 | 0.1299 | 0.2028 | 0.0384 |
92
+ | 0.8667 | 28.94 | 9000 | 0.1228 | 0.1945 | 0.0367 |
93
+ | 0.8641 | 30.55 | 9500 | 0.1223 | 0.1939 | 0.0364 |
94
+ | 0.8516 | 32.15 | 10000 | 0.1184 | 0.1876 | 0.0349 |
95
+ | 0.8379 | 33.76 | 10500 | 0.1137 | 0.1821 | 0.0338 |
96
+ | 0.8235 | 35.37 | 11000 | 0.1127 | 0.1779 | 0.0331 |
97
+ | 0.8112 | 36.98 | 11500 | 0.1103 | 0.1766 | 0.0327 |
98
+ | 0.8069 | 38.59 | 12000 | 0.1092 | 0.1752 | 0.0323 |
99
 
100
 
101
  ### Framework versions
log_speech-recognition-community-v2_dev_data_uk_validation_predictions.txt ADDED
The diff for this file is too large to render. See raw diff
 
log_speech-recognition-community-v2_dev_data_uk_validation_targets.txt ADDED
The diff for this file is too large to render. See raw diff