Rastadayon
commited on
Commit
•
50b2d5d
1
Parent(s):
4aefbcc
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
datasets:
|
6 |
+
- common_voice
|
7 |
+
model-index:
|
8 |
+
- name: wav2vec2-large-xls-r-300m-dutch-baseline
|
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 |
+
# wav2vec2-large-xls-r-300m-dutch-baseline
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 0.5107
|
20 |
+
- Wer: 0.2674
|
21 |
+
- Cer: 0.0863
|
22 |
+
|
23 |
+
## Model description
|
24 |
+
|
25 |
+
More information needed
|
26 |
+
|
27 |
+
## Intended uses & limitations
|
28 |
+
|
29 |
+
More information needed
|
30 |
+
|
31 |
+
## Training and evaluation data
|
32 |
+
|
33 |
+
More information needed
|
34 |
+
|
35 |
+
## Training procedure
|
36 |
+
|
37 |
+
### Training hyperparameters
|
38 |
+
|
39 |
+
The following hyperparameters were used during training:
|
40 |
+
- learning_rate: 0.0003
|
41 |
+
- train_batch_size: 4
|
42 |
+
- eval_batch_size: 4
|
43 |
+
- seed: 42
|
44 |
+
- distributed_type: multi-GPU
|
45 |
+
- num_devices: 2
|
46 |
+
- gradient_accumulation_steps: 4
|
47 |
+
- total_train_batch_size: 32
|
48 |
+
- total_eval_batch_size: 8
|
49 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
50 |
+
- lr_scheduler_type: linear
|
51 |
+
- lr_scheduler_warmup_steps: 500
|
52 |
+
- num_epochs: 30
|
53 |
+
- mixed_precision_training: Native AMP
|
54 |
+
|
55 |
+
### Training results
|
56 |
+
|
57 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
58 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
59 |
+
| 3.655 | 1.31 | 400 | 0.9337 | 0.7332 | 0.2534 |
|
60 |
+
| 0.42 | 2.61 | 800 | 0.5018 | 0.4115 | 0.1374 |
|
61 |
+
| 0.2267 | 3.92 | 1200 | 0.4776 | 0.3791 | 0.1259 |
|
62 |
+
| 0.1624 | 5.23 | 1600 | 0.4807 | 0.3590 | 0.1208 |
|
63 |
+
| 0.135 | 6.54 | 2000 | 0.4899 | 0.3417 | 0.1121 |
|
64 |
+
| 0.1179 | 7.84 | 2400 | 0.5096 | 0.3445 | 0.1133 |
|
65 |
+
| 0.1035 | 9.15 | 2800 | 0.4563 | 0.3455 | 0.1129 |
|
66 |
+
| 0.092 | 10.46 | 3200 | 0.5061 | 0.3382 | 0.1127 |
|
67 |
+
| 0.0804 | 11.76 | 3600 | 0.4969 | 0.3285 | 0.1088 |
|
68 |
+
| 0.0748 | 13.07 | 4000 | 0.5274 | 0.3380 | 0.1114 |
|
69 |
+
| 0.0669 | 14.38 | 4400 | 0.5201 | 0.3115 | 0.1028 |
|
70 |
+
| 0.0588 | 15.69 | 4800 | 0.5238 | 0.3212 | 0.1054 |
|
71 |
+
| 0.0561 | 16.99 | 5200 | 0.5273 | 0.3185 | 0.1044 |
|
72 |
+
| 0.0513 | 18.3 | 5600 | 0.5577 | 0.3032 | 0.1010 |
|
73 |
+
| 0.0476 | 19.61 | 6000 | 0.5298 | 0.3050 | 0.1008 |
|
74 |
+
| 0.0408 | 20.91 | 6400 | 0.5725 | 0.2982 | 0.0984 |
|
75 |
+
| 0.0376 | 22.22 | 6800 | 0.5605 | 0.2953 | 0.0966 |
|
76 |
+
| 0.0339 | 23.53 | 7200 | 0.5419 | 0.2865 | 0.0938 |
|
77 |
+
| 0.0315 | 24.84 | 7600 | 0.5530 | 0.2782 | 0.0915 |
|
78 |
+
| 0.0286 | 26.14 | 8000 | 0.5354 | 0.2788 | 0.0917 |
|
79 |
+
| 0.0259 | 27.45 | 8400 | 0.5245 | 0.2715 | 0.0878 |
|
80 |
+
| 0.0231 | 28.76 | 8800 | 0.5107 | 0.2674 | 0.0863 |
|
81 |
+
|
82 |
+
|
83 |
+
### Framework versions
|
84 |
+
|
85 |
+
- Transformers 4.17.0
|
86 |
+
- Pytorch 1.12.0+cu102
|
87 |
+
- Datasets 2.7.1
|
88 |
+
- Tokenizers 0.13.2
|