đ xtreme_s_xlsr_300m_minds14
This model is a fine - tuned version of facebook/wav2vec2-xls-r-300m on the GOOGLE/XTREME_S - MINDS14.ALL dataset. It offers high - performance evaluation results across multiple languages and metrics, providing valuable insights for speech - related tasks.
đ Quick Start
This model is ready to use for speech - related tasks. You can directly load it from Hugging Face and start making inferences.
đ Documentation
Model Evaluation Results
It achieves the following results on the evaluation set:
- Accuracy: 0.9033
- Accuracy Cs - cz: 0.9164
- Accuracy De - de: 0.9477
- Accuracy En - au: 0.9235
- Accuracy En - gb: 0.9324
- Accuracy En - us: 0.9326
- Accuracy Es - es: 0.9177
- Accuracy Fr - fr: 0.9444
- Accuracy It - it: 0.9167
- Accuracy Ko - kr: 0.8649
- Accuracy Nl - nl: 0.9450
- Accuracy Pl - pl: 0.9146
- Accuracy Pt - pt: 0.8940
- Accuracy Ru - ru: 0.8667
- Accuracy Zh - cn: 0.7291
- F1: 0.9015
- F1 Cs - cz: 0.9154
- F1 De - de: 0.9467
- F1 En - au: 0.9199
- F1 En - gb: 0.9334
- F1 En - us: 0.9308
- F1 Es - es: 0.9158
- F1 Fr - fr: 0.9436
- F1 It - it: 0.9135
- F1 Ko - kr: 0.8642
- F1 Nl - nl: 0.9440
- F1 Pl - pl: 0.9159
- F1 Pt - pt: 0.8883
- F1 Ru - ru: 0.8646
- F1 Zh - cn: 0.7249
- Loss: 0.4119
- Loss Cs - cz: 0.3790
- Loss De - de: 0.2649
- Loss En - au: 0.3459
- Loss En - gb: 0.2853
- Loss En - us: 0.2203
- Loss Es - es: 0.2731
- Loss Fr - fr: 0.1909
- Loss It - it: 0.3520
- Loss Ko - kr: 0.5431
- Loss Nl - nl: 0.2515
- Loss Pl - pl: 0.4113
- Loss Pt - pt: 0.4798
- Loss Ru - ru: 0.6470
- Loss Zh - cn: 1.1216
- Predict Samples: 4086
Training Procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- distributed_type: multi - GPU
- num_devices: 2
- total_train_batch_size: 64
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1500
- num_epochs: 50.0
- mixed_precision_training: Native AMP
Training results
| Training Loss |
Epoch |
Step |
Validation Loss |
F1 |
Accuracy |
| 2.6739 |
5.41 |
200 |
2.5687 |
0.0430 |
0.1190 |
| 1.4953 |
10.81 |
400 |
1.6052 |
0.5550 |
0.5692 |
| 0.6177 |
16.22 |
600 |
0.7927 |
0.8052 |
0.8011 |
| 0.3609 |
21.62 |
800 |
0.5679 |
0.8609 |
0.8609 |
| 0.4972 |
27.03 |
1000 |
0.5944 |
0.8509 |
0.8523 |
| 0.1799 |
32.43 |
1200 |
0.6194 |
0.8623 |
0.8621 |
| 0.1308 |
37.84 |
1400 |
0.5956 |
0.8569 |
0.8548 |
| 0.2298 |
43.24 |
1600 |
0.5201 |
0.8732 |
0.8743 |
| 0.0052 |
48.65 |
1800 |
0.3826 |
0.9106 |
0.9103 |
Framework versions
- Transformers 4.18.0.dev0
- Pytorch 1.10.2+cu113
- Datasets 2.0.1.dev0
- Tokenizers 0.11.6
đ License
This project is licensed under the Apache - 2.0 license.
| Property |
Details |
| Model Type |
Fine - tuned version of facebook/wav2vec2 - xls - r - 300m on GOOGLE/XTREME_S - MINDS14.ALL dataset |
| Training Data |
GOOGLE/XTREME_S - MINDS14.ALL dataset |
| Metrics |
F1, Accuracy |
%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
PyTorch