File size: 1,686 Bytes
3951734 b4dfbf0 3951734 b4dfbf0 3951734 d1f5822 3951734 948baeb d1f5822 3951734 948baeb 3951734 948baeb 3951734 948baeb 3951734 948baeb 3951734 948baeb 3951734 948baeb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 |
---
language:
- uz
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_10_0
- generated_from_trainer
- AIRI_UZ
datasets:
- common_voice_10_0
model-index:
- name: uzbek_stt
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
## Oyqiz jamoasi a'zolari tomonidan qilingan STT ning eng kichik versiyasi. Bu yaxshi versiya emas faqat 15ta epoch o'qitilgan!
### Foziljon To'lqinov, Shaxboz Zohidov, Abduraxim Jabborov, Yahyoxon Rahimov
Bu model [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) va MOZILLA-FOUNDATION/COMMON_VOICE_10_0 - UZ dataseti bilan 15ta epoxta o'qitilgan.
O'qitish natijalari:
- Xatolik: 0.5763
- So'z xatoligi: 0.4502
### O'qitish giperparameterlari
O'qitish uchun ishlatilgan giperparameterlar:
- learning_rate: 0.0003
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- total_train_batch_size: 8
- 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: 500
- num_epochs: 15.0
- mixed_precision_training: Native AMP
### O'qitish natijalari
| Xatolik | Epox | Qadam | Tasdiq xatoligi | SXD |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 0.4736 | 3.4 | 10000 | 0.6247 | 0.6362 |
| 0.3392 | 6.8 | 20000 | 0.7254 | 0.5605 |
| 0.2085 | 10.19 | 30000 | 0.5465 | 0.5097 |
| 0.1387 | 13.59 | 40000 | 0.5984 | 0.4632 | |