lucas-meyer
commited on
Commit
•
2237733
1
Parent(s):
6fcac5a
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
metrics:
|
6 |
+
- wer
|
7 |
+
model-index:
|
8 |
+
- name: xls-r-asr_af-run1
|
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 |
+
# xls-r-asr_af-run1
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on an unknown dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 0.4789
|
20 |
+
- Wer: 0.4201
|
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: 0.0003
|
40 |
+
- train_batch_size: 4
|
41 |
+
- eval_batch_size: 4
|
42 |
+
- seed: 42
|
43 |
+
- gradient_accumulation_steps: 3
|
44 |
+
- total_train_batch_size: 12
|
45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
+
- lr_scheduler_type: linear
|
47 |
+
- lr_scheduler_warmup_steps: 500
|
48 |
+
- num_epochs: 30
|
49 |
+
- mixed_precision_training: Native AMP
|
50 |
+
|
51 |
+
### Training results
|
52 |
+
|
53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
54 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
55 |
+
| 11.8837 | 0.22 | 50 | 7.3321 | 1.0 |
|
56 |
+
| 4.9879 | 0.44 | 100 | 4.0442 | 1.0 |
|
57 |
+
| 3.5299 | 0.66 | 150 | 3.1941 | 1.0 |
|
58 |
+
| 3.0621 | 0.88 | 200 | 2.9995 | 1.0 |
|
59 |
+
| 2.9941 | 1.1 | 250 | 3.0224 | 1.0 |
|
60 |
+
| 2.9668 | 1.32 | 300 | 2.9470 | 1.0 |
|
61 |
+
| 2.9351 | 1.54 | 350 | 2.9053 | 1.0 |
|
62 |
+
| 2.5839 | 1.76 | 400 | 1.9778 | 0.9929 |
|
63 |
+
| 1.7074 | 1.98 | 450 | 1.3715 | 0.8741 |
|
64 |
+
| 1.2129 | 2.2 | 500 | 1.2575 | 0.9809 |
|
65 |
+
| 1.0397 | 2.42 | 550 | 1.0340 | 0.7792 |
|
66 |
+
| 0.9081 | 2.64 | 600 | 0.9177 | 0.7051 |
|
67 |
+
| 0.7855 | 2.86 | 650 | 0.7769 | 0.6323 |
|
68 |
+
| 0.7099 | 3.08 | 700 | 0.6570 | 0.5771 |
|
69 |
+
| 0.5658 | 3.3 | 750 | 0.6074 | 0.5456 |
|
70 |
+
| 0.5115 | 3.52 | 800 | 0.5645 | 0.5291 |
|
71 |
+
| 0.5005 | 3.74 | 850 | 0.5539 | 0.5124 |
|
72 |
+
| 0.4892 | 3.96 | 900 | 0.5390 | 0.5105 |
|
73 |
+
| 0.3787 | 4.19 | 950 | 0.5359 | 0.4879 |
|
74 |
+
| 0.3559 | 4.41 | 1000 | 0.5400 | 0.4904 |
|
75 |
+
| 0.3349 | 4.63 | 1050 | 0.5019 | 0.4672 |
|
76 |
+
| 0.3374 | 4.85 | 1100 | 0.4940 | 0.4519 |
|
77 |
+
| 0.3423 | 5.07 | 1150 | 0.5030 | 0.4364 |
|
78 |
+
| 0.257 | 5.29 | 1200 | 0.4646 | 0.4243 |
|
79 |
+
| 0.2431 | 5.51 | 1250 | 0.4803 | 0.4330 |
|
80 |
+
| 0.2598 | 5.73 | 1300 | 0.5143 | 0.4398 |
|
81 |
+
| 0.2602 | 5.95 | 1350 | 0.4789 | 0.4201 |
|
82 |
+
|
83 |
+
|
84 |
+
### Framework versions
|
85 |
+
|
86 |
+
- Transformers 4.28.0
|
87 |
+
- Pytorch 2.0.1+cu117
|
88 |
+
- Datasets 2.14.4
|
89 |
+
- Tokenizers 0.13.3
|