Commit
•
c9aff04
1
Parent(s):
1275d57
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,95 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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-spanish-large
|
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-spanish-large
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [tomascufaro/xls-r-es-test](https://huggingface.co/tomascufaro/xls-r-es-test) on the common_voice dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 0.1431
|
20 |
+
- Wer: 0.1197
|
21 |
+
|
22 |
+
## Model description
|
23 |
+
|
24 |
+
More information needed
|
25 |
+
|
26 |
+
## Intended uses & limitations
|
27 |
+
|
28 |
+
More information needed
|
29 |
+
|
30 |
+
## Training and evaluation data
|
31 |
+
|
32 |
+
More information needed
|
33 |
+
|
34 |
+
## Training procedure
|
35 |
+
|
36 |
+
### Training hyperparameters
|
37 |
+
|
38 |
+
The following hyperparameters were used during training:
|
39 |
+
- learning_rate: 0.0002
|
40 |
+
- train_batch_size: 10
|
41 |
+
- eval_batch_size: 8
|
42 |
+
- seed: 42
|
43 |
+
- gradient_accumulation_steps: 2
|
44 |
+
- total_train_batch_size: 20
|
45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
+
- lr_scheduler_type: linear
|
47 |
+
- lr_scheduler_warmup_steps: 300
|
48 |
+
- num_epochs: 5
|
49 |
+
- mixed_precision_training: Native AMP
|
50 |
+
|
51 |
+
### Training results
|
52 |
+
|
53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
54 |
+
|:-------------:|:-----:|:-----:|:---------------:|:------:|
|
55 |
+
| 0.1769 | 0.15 | 400 | 0.1795 | 0.1698 |
|
56 |
+
| 0.217 | 0.3 | 800 | 0.2000 | 0.1945 |
|
57 |
+
| 0.2372 | 0.45 | 1200 | 0.1985 | 0.1859 |
|
58 |
+
| 0.2351 | 0.6 | 1600 | 0.1901 | 0.1772 |
|
59 |
+
| 0.2269 | 0.75 | 2000 | 0.1968 | 0.1783 |
|
60 |
+
| 0.2284 | 0.9 | 2400 | 0.1873 | 0.1771 |
|
61 |
+
| 0.2014 | 1.06 | 2800 | 0.1840 | 0.1696 |
|
62 |
+
| 0.1988 | 1.21 | 3200 | 0.1904 | 0.1730 |
|
63 |
+
| 0.1919 | 1.36 | 3600 | 0.1827 | 0.1630 |
|
64 |
+
| 0.1919 | 1.51 | 4000 | 0.1788 | 0.1629 |
|
65 |
+
| 0.1817 | 1.66 | 4400 | 0.1755 | 0.1558 |
|
66 |
+
| 0.1812 | 1.81 | 4800 | 0.1795 | 0.1638 |
|
67 |
+
| 0.1808 | 1.96 | 5200 | 0.1762 | 0.1603 |
|
68 |
+
| 0.1625 | 2.11 | 5600 | 0.1721 | 0.1557 |
|
69 |
+
| 0.1477 | 2.26 | 6000 | 0.1735 | 0.1504 |
|
70 |
+
| 0.1508 | 2.41 | 6400 | 0.1708 | 0.1478 |
|
71 |
+
| 0.157 | 2.56 | 6800 | 0.1644 | 0.1466 |
|
72 |
+
| 0.1491 | 2.71 | 7200 | 0.1638 | 0.1445 |
|
73 |
+
| 0.1458 | 2.86 | 7600 | 0.1582 | 0.1426 |
|
74 |
+
| 0.1387 | 3.02 | 8000 | 0.1607 | 0.1376 |
|
75 |
+
| 0.1269 | 3.17 | 8400 | 0.1559 | 0.1364 |
|
76 |
+
| 0.1172 | 3.32 | 8800 | 0.1521 | 0.1335 |
|
77 |
+
| 0.1203 | 3.47 | 9200 | 0.1534 | 0.1330 |
|
78 |
+
| 0.1177 | 3.62 | 9600 | 0.1485 | 0.1304 |
|
79 |
+
| 0.1167 | 3.77 | 10000 | 0.1498 | 0.1302 |
|
80 |
+
| 0.1194 | 3.92 | 10400 | 0.1463 | 0.1287 |
|
81 |
+
| 0.1053 | 4.07 | 10800 | 0.1483 | 0.1282 |
|
82 |
+
| 0.098 | 4.22 | 11200 | 0.1498 | 0.1267 |
|
83 |
+
| 0.0958 | 4.37 | 11600 | 0.1461 | 0.1233 |
|
84 |
+
| 0.0946 | 4.52 | 12000 | 0.1444 | 0.1218 |
|
85 |
+
| 0.094 | 4.67 | 12400 | 0.1434 | 0.1206 |
|
86 |
+
| 0.0932 | 4.82 | 12800 | 0.1424 | 0.1206 |
|
87 |
+
| 0.0912 | 4.98 | 13200 | 0.1431 | 0.1197 |
|
88 |
+
|
89 |
+
|
90 |
+
### Framework versions
|
91 |
+
|
92 |
+
- Transformers 4.17.0.dev0
|
93 |
+
- Pytorch 1.10.2+cu102
|
94 |
+
- Datasets 1.18.2.dev0
|
95 |
+
- Tokenizers 0.11.0
|