StevenLimcorn
commited on
Commit
·
a214e6b
1
Parent(s):
97bab54
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
datasets:
|
6 |
+
- common_voice
|
7 |
+
model-index:
|
8 |
+
- name: ''
|
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 |
+
#
|
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: 1.1801
|
20 |
+
- Wer: 0.8621
|
21 |
+
- Cer: 0.2969
|
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: 7.5e-05
|
41 |
+
- train_batch_size: 8
|
42 |
+
- eval_batch_size: 8
|
43 |
+
- seed: 42
|
44 |
+
- gradient_accumulation_steps: 4
|
45 |
+
- total_train_batch_size: 32
|
46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
47 |
+
- lr_scheduler_type: linear
|
48 |
+
- lr_scheduler_warmup_steps: 2000
|
49 |
+
- num_epochs: 100.0
|
50 |
+
- mixed_precision_training: Native AMP
|
51 |
+
|
52 |
+
### Training results
|
53 |
+
|
54 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
55 |
+
|:-------------:|:-----:|:-----:|:---------------:|:------:|:------:|
|
56 |
+
| 64.6189 | 2.51 | 500 | 63.8077 | 1.0 | 1.0 |
|
57 |
+
| 8.0561 | 5.03 | 1000 | 6.8014 | 1.0 | 1.0 |
|
58 |
+
| 6.0427 | 7.54 | 1500 | 6.0745 | 1.0 | 1.0 |
|
59 |
+
| 5.9357 | 10.05 | 2000 | 5.8682 | 1.0 | 1.0 |
|
60 |
+
| 5.0489 | 12.56 | 2500 | 4.4032 | 0.9990 | 0.7750 |
|
61 |
+
| 4.6184 | 15.08 | 3000 | 3.8383 | 0.9983 | 0.6768 |
|
62 |
+
| 4.365 | 17.59 | 3500 | 3.4633 | 0.9959 | 0.6299 |
|
63 |
+
| 4.1026 | 20.1 | 4000 | 3.0732 | 0.9902 | 0.5814 |
|
64 |
+
| 3.8655 | 22.61 | 4500 | 2.7638 | 0.9868 | 0.5465 |
|
65 |
+
| 3.6991 | 25.13 | 5000 | 2.4759 | 0.9811 | 0.5088 |
|
66 |
+
| 3.4894 | 27.64 | 5500 | 2.2937 | 0.9746 | 0.4852 |
|
67 |
+
| 3.3983 | 30.15 | 6000 | 2.1684 | 0.9733 | 0.4674 |
|
68 |
+
| 3.2736 | 32.66 | 6500 | 2.0372 | 0.9659 | 0.4458 |
|
69 |
+
| 3.1884 | 35.18 | 7000 | 1.9267 | 0.9648 | 0.4329 |
|
70 |
+
| 3.1248 | 37.69 | 7500 | 1.8408 | 0.9591 | 0.4217 |
|
71 |
+
| 3.0381 | 40.2 | 8000 | 1.7531 | 0.9503 | 0.4074 |
|
72 |
+
| 2.9515 | 42.71 | 8500 | 1.6880 | 0.9459 | 0.3967 |
|
73 |
+
| 2.8704 | 45.23 | 9000 | 1.6264 | 0.9378 | 0.3884 |
|
74 |
+
| 2.8128 | 47.74 | 9500 | 1.5621 | 0.9341 | 0.3782 |
|
75 |
+
| 2.7386 | 50.25 | 10000 | 1.5011 | 0.9243 | 0.3664 |
|
76 |
+
| 2.6646 | 52.76 | 10500 | 1.4608 | 0.9192 | 0.3575 |
|
77 |
+
| 2.6072 | 55.28 | 11000 | 1.4251 | 0.9148 | 0.3501 |
|
78 |
+
| 2.569 | 57.79 | 11500 | 1.3837 | 0.9060 | 0.3462 |
|
79 |
+
| 2.5091 | 60.3 | 12000 | 1.3589 | 0.9070 | 0.3392 |
|
80 |
+
| 2.4588 | 62.81 | 12500 | 1.3261 | 0.8966 | 0.3284 |
|
81 |
+
| 2.4083 | 65.33 | 13000 | 1.3052 | 0.8982 | 0.3265 |
|
82 |
+
| 2.3787 | 67.84 | 13500 | 1.2997 | 0.8908 | 0.3243 |
|
83 |
+
| 2.3457 | 70.35 | 14000 | 1.2778 | 0.8898 | 0.3187 |
|
84 |
+
| 2.3099 | 72.86 | 14500 | 1.2661 | 0.8830 | 0.3172 |
|
85 |
+
| 2.2559 | 75.38 | 15000 | 1.2475 | 0.8851 | 0.3143 |
|
86 |
+
| 2.2264 | 77.89 | 15500 | 1.2319 | 0.8739 | 0.3085 |
|
87 |
+
| 2.196 | 80.4 | 16000 | 1.2218 | 0.8722 | 0.3049 |
|
88 |
+
| 2.1613 | 82.91 | 16500 | 1.2093 | 0.8719 | 0.3051 |
|
89 |
+
| 2.1455 | 85.43 | 17000 | 1.2055 | 0.8624 | 0.3005 |
|
90 |
+
| 2.1193 | 87.94 | 17500 | 1.1975 | 0.8600 | 0.2982 |
|
91 |
+
| 2.0911 | 90.45 | 18000 | 1.1960 | 0.8648 | 0.3003 |
|
92 |
+
| 2.0884 | 92.96 | 18500 | 1.1871 | 0.8638 | 0.2971 |
|
93 |
+
| 2.0766 | 95.48 | 19000 | 1.1814 | 0.8617 | 0.2967 |
|
94 |
+
| 2.0735 | 97.99 | 19500 | 1.1801 | 0.8621 | 0.2969 |
|
95 |
+
|
96 |
+
|
97 |
+
### Framework versions
|
98 |
+
|
99 |
+
- Transformers 4.17.0.dev0
|
100 |
+
- Pytorch 1.10.2+cu102
|
101 |
+
- Datasets 1.18.2.dev0
|
102 |
+
- Tokenizers 0.11.0
|
nohup.out
CHANGED
@@ -15688,3 +15688,22 @@ Saving model checkpoint to ./
|
|
15688 |
Configuration saved in ./config.json
|
15689 |
Model weights saved in ./pytorch_model.bin
|
15690 |
Configuration saved in ./preprocessor_config.json
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15688 |
Configuration saved in ./config.json
|
15689 |
Model weights saved in ./pytorch_model.bin
|
15690 |
Configuration saved in ./preprocessor_config.json
|
15691 |
+
{'eval_loss': 1.180108904838562, 'eval_wer': 0.8620689655172413, 'eval_cer': 0.2968590450516685, 'eval_runtime': 123.9462, 'eval_samples_per_second': 23.865, 'eval_steps_per_second': 2.985, 'epoch': 97.99}
|
15692 |
+
{'loss': 2.0277, 'learning_rate': 1.3156424581005587e-06, 'epoch': 98.49}
|
15693 |
+
{'loss': 2.0545, 'learning_rate': 8.966480446927373e-07, 'epoch': 98.99}
|
15694 |
+
{'loss': 2.0502, 'learning_rate': 4.776536312849162e-07, 'epoch': 99.5}
|
15695 |
+
{'loss': 2.0333, 'learning_rate': 5.865921787709496e-08, 'epoch': 100.0}
|
15696 |
+
{'train_runtime': 36587.1788, 'train_samples_per_second': 17.438, 'train_steps_per_second': 0.544, 'train_loss': 5.91964619928868, 'epoch': 100.0}
|
15697 |
+
|
15698 |
+
|
15699 |
+
b94fb3e..97bab54 main -> main
|
15700 |
+
|
15701 |
+
02/06/2022 21:54:30 - WARNING - huggingface_hub.repository - To https://huggingface.co/StevenLimcorn/wav2vec2-xls-r-300m-zh-TW
|
15702 |
+
b94fb3e..97bab54 main -> main
|
15703 |
+
|
15704 |
+
|
15705 |
+
|
15706 |
+
|
15707 |
+
|
15708 |
+
Dropping the following result as it does not have all the necessary fields:
|
15709 |
+
{'dataset': {'name': 'common_voice', 'type': 'common_voice', 'args': 'zh-TW'}}
|