samil24 commited on
Commit
a40fb74
·
verified ·
1 Parent(s): e26fc8c

End of training

Browse files
Files changed (1) hide show
  1. README.md +106 -0
README.md ADDED
@@ -0,0 +1,106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ base_model: facebook/wav2vec2-large-xlsr-53
5
+ tags:
6
+ - generated_from_trainer
7
+ metrics:
8
+ - wer
9
+ model-index:
10
+ - name: wav2vec2-large-xlsr-sorani-v1
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # wav2vec2-large-xlsr-sorani-v1
18
+
19
+ This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the None dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.2492
22
+ - Wer: 0.3566
23
+
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 0.0001
42
+ - train_batch_size: 16
43
+ - eval_batch_size: 8
44
+ - seed: 42
45
+ - gradient_accumulation_steps: 2
46
+ - total_train_batch_size: 32
47
+ - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
48
+ - lr_scheduler_type: linear
49
+ - lr_scheduler_warmup_steps: 1000
50
+ - num_epochs: 15
51
+ - mixed_precision_training: Native AMP
52
+
53
+ ### Training results
54
+
55
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
56
+ |:-------------:|:-------:|:----:|:---------------:|:------:|
57
+ | 9.2607 | 0.3543 | 200 | 6.4854 | 1.0 |
58
+ | 3.0716 | 0.7086 | 400 | 3.0408 | 1.0 |
59
+ | 2.8638 | 1.0620 | 600 | 2.8982 | 1.0 |
60
+ | 2.8666 | 1.4163 | 800 | 2.8511 | 1.0 |
61
+ | 1.8111 | 1.7706 | 1000 | 1.5181 | 1.0095 |
62
+ | 0.7703 | 2.1240 | 1200 | 0.5528 | 0.6563 |
63
+ | 0.6864 | 2.4783 | 1400 | 0.4600 | 0.5909 |
64
+ | 0.5955 | 2.8326 | 1600 | 0.4036 | 0.5362 |
65
+ | 0.4894 | 3.1860 | 1800 | 0.3811 | 0.5030 |
66
+ | 0.5007 | 3.5403 | 2000 | 0.3551 | 0.4782 |
67
+ | 0.5169 | 3.8946 | 2200 | 0.3340 | 0.4627 |
68
+ | 0.4613 | 4.2480 | 2400 | 0.3271 | 0.4527 |
69
+ | 0.4499 | 4.6023 | 2600 | 0.3229 | 0.4437 |
70
+ | 0.4093 | 4.9566 | 2800 | 0.3042 | 0.4333 |
71
+ | 0.4458 | 5.3100 | 3000 | 0.3040 | 0.4316 |
72
+ | 0.4107 | 5.6643 | 3200 | 0.2875 | 0.4162 |
73
+ | 0.3728 | 6.0177 | 3400 | 0.2862 | 0.4150 |
74
+ | 0.348 | 6.3720 | 3600 | 0.2939 | 0.4068 |
75
+ | 0.3393 | 6.7263 | 3800 | 0.2755 | 0.3977 |
76
+ | 0.3572 | 7.0797 | 4000 | 0.2799 | 0.3975 |
77
+ | 0.3453 | 7.4340 | 4200 | 0.2748 | 0.3975 |
78
+ | 0.3434 | 7.7883 | 4400 | 0.2688 | 0.3907 |
79
+ | 0.3231 | 8.1417 | 4600 | 0.2706 | 0.3871 |
80
+ | 0.3148 | 8.4960 | 4800 | 0.2688 | 0.3830 |
81
+ | 0.3261 | 8.8503 | 5000 | 0.2625 | 0.3811 |
82
+ | 0.3055 | 9.2037 | 5200 | 0.2597 | 0.3719 |
83
+ | 0.3028 | 9.5580 | 5400 | 0.2593 | 0.3720 |
84
+ | 0.3246 | 9.9123 | 5600 | 0.2535 | 0.3724 |
85
+ | 0.274 | 10.2657 | 5800 | 0.2551 | 0.3690 |
86
+ | 0.2832 | 10.6200 | 6000 | 0.2601 | 0.3716 |
87
+ | 0.2899 | 10.9743 | 6200 | 0.2521 | 0.3670 |
88
+ | 0.2889 | 11.3277 | 6400 | 0.2574 | 0.3652 |
89
+ | 0.2675 | 11.6820 | 6600 | 0.2564 | 0.3611 |
90
+ | 0.3334 | 12.0354 | 6800 | 0.2548 | 0.3603 |
91
+ | 0.2568 | 12.3897 | 7000 | 0.2538 | 0.3606 |
92
+ | 0.2842 | 12.7440 | 7200 | 0.2566 | 0.3597 |
93
+ | 0.2631 | 13.0974 | 7400 | 0.2540 | 0.3580 |
94
+ | 0.2573 | 13.4517 | 7600 | 0.2513 | 0.3569 |
95
+ | 0.27 | 13.8060 | 7800 | 0.2499 | 0.3556 |
96
+ | 0.2578 | 14.1594 | 8000 | 0.2496 | 0.3556 |
97
+ | 0.2528 | 14.5137 | 8200 | 0.2498 | 0.3565 |
98
+ | 0.2644 | 14.8680 | 8400 | 0.2492 | 0.3566 |
99
+
100
+
101
+ ### Framework versions
102
+
103
+ - Transformers 4.54.1
104
+ - Pytorch 2.7.1+cu128
105
+ - Datasets 3.6.0
106
+ - Tokenizers 0.21.4