update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
model-index:
|
6 |
+
- name: korean-aihub-learning-math-1-test
|
7 |
+
results: []
|
8 |
+
---
|
9 |
+
|
10 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
11 |
+
should probably proofread and complete it, then remove this comment. -->
|
12 |
+
|
13 |
+
# korean-aihub-learning-math-1-test
|
14 |
+
|
15 |
+
This model is a fine-tuned version of [kresnik/wav2vec2-large-xlsr-korean](https://huggingface.co/kresnik/wav2vec2-large-xlsr-korean) on the None dataset.
|
16 |
+
It achieves the following results on the evaluation set:
|
17 |
+
- Loss: 1.2537
|
18 |
+
- Wer: 0.4765
|
19 |
+
|
20 |
+
## Model description
|
21 |
+
|
22 |
+
More information needed
|
23 |
+
|
24 |
+
## Intended uses & limitations
|
25 |
+
|
26 |
+
More information needed
|
27 |
+
|
28 |
+
## Training and evaluation data
|
29 |
+
|
30 |
+
More information needed
|
31 |
+
|
32 |
+
## Training procedure
|
33 |
+
|
34 |
+
### Training hyperparameters
|
35 |
+
|
36 |
+
The following hyperparameters were used during training:
|
37 |
+
- learning_rate: 0.0003
|
38 |
+
- train_batch_size: 4
|
39 |
+
- eval_batch_size: 8
|
40 |
+
- seed: 42
|
41 |
+
- gradient_accumulation_steps: 2
|
42 |
+
- total_train_batch_size: 8
|
43 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
44 |
+
- lr_scheduler_type: linear
|
45 |
+
- lr_scheduler_warmup_steps: 500
|
46 |
+
- num_epochs: 30
|
47 |
+
- mixed_precision_training: Native AMP
|
48 |
+
|
49 |
+
### Training results
|
50 |
+
|
51 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
52 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
53 |
+
| No log | 1.0 | 35 | 29.8031 | 1.0 |
|
54 |
+
| No log | 2.0 | 70 | 5.7158 | 1.0 |
|
55 |
+
| 19.8789 | 3.0 | 105 | 4.5005 | 1.0 |
|
56 |
+
| 19.8789 | 4.0 | 140 | 4.3677 | 0.9984 |
|
57 |
+
| 19.8789 | 5.0 | 175 | 3.8013 | 0.9882 |
|
58 |
+
| 3.9785 | 6.0 | 210 | 2.4132 | 0.8730 |
|
59 |
+
| 3.9785 | 7.0 | 245 | 1.5867 | 0.7045 |
|
60 |
+
| 3.9785 | 8.0 | 280 | 1.3179 | 0.6082 |
|
61 |
+
| 1.2266 | 9.0 | 315 | 1.2431 | 0.6066 |
|
62 |
+
| 1.2266 | 10.0 | 350 | 1.1791 | 0.5384 |
|
63 |
+
| 1.2266 | 11.0 | 385 | 1.0994 | 0.5298 |
|
64 |
+
| 0.3916 | 12.0 | 420 | 1.1552 | 0.5196 |
|
65 |
+
| 0.3916 | 13.0 | 455 | 1.1495 | 0.5486 |
|
66 |
+
| 0.3916 | 14.0 | 490 | 1.1340 | 0.5290 |
|
67 |
+
| 0.2488 | 15.0 | 525 | 1.2208 | 0.5525 |
|
68 |
+
| 0.2488 | 16.0 | 560 | 1.1682 | 0.5024 |
|
69 |
+
| 0.2488 | 17.0 | 595 | 1.1479 | 0.5008 |
|
70 |
+
| 0.1907 | 18.0 | 630 | 1.1735 | 0.4882 |
|
71 |
+
| 0.1907 | 19.0 | 665 | 1.2302 | 0.4914 |
|
72 |
+
| 0.1461 | 20.0 | 700 | 1.2497 | 0.4890 |
|
73 |
+
| 0.1461 | 21.0 | 735 | 1.2434 | 0.4914 |
|
74 |
+
| 0.1461 | 22.0 | 770 | 1.2031 | 0.5031 |
|
75 |
+
| 0.1147 | 23.0 | 805 | 1.2451 | 0.4976 |
|
76 |
+
| 0.1147 | 24.0 | 840 | 1.2746 | 0.4937 |
|
77 |
+
| 0.1147 | 25.0 | 875 | 1.2405 | 0.4828 |
|
78 |
+
| 0.0892 | 26.0 | 910 | 1.2228 | 0.4929 |
|
79 |
+
| 0.0892 | 27.0 | 945 | 1.2642 | 0.4898 |
|
80 |
+
| 0.0892 | 28.0 | 980 | 1.2586 | 0.4843 |
|
81 |
+
| 0.0709 | 29.0 | 1015 | 1.2518 | 0.4788 |
|
82 |
+
| 0.0709 | 30.0 | 1050 | 1.2537 | 0.4765 |
|
83 |
+
|
84 |
+
|
85 |
+
### Framework versions
|
86 |
+
|
87 |
+
- Transformers 4.20.1
|
88 |
+
- Pytorch 1.12.0+cu113
|
89 |
+
- Datasets 2.4.0
|
90 |
+
- Tokenizers 0.12.1
|