File size: 1,746 Bytes
bc32137 |
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 56 57 58 59 60 61 62 63 64 65 66 67 68 |
---
license: apache-2.0
base_model: jonatasgrosman/wav2vec2-large-xlsr-53-arabic
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: tamasheq-98
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. -->
# tamasheq-98
This model is a fine-tuned version of [jonatasgrosman/wav2vec2-large-xlsr-53-arabic](https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-arabic) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.5880
- Wer: 0.9741
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 11.747 | 6.0 | 300 | 3.0325 | 1.0 |
| 2.6986 | 12.0 | 600 | 2.2258 | 1.0148 |
| 0.5256 | 18.0 | 900 | 1.7011 | 0.9852 |
| 0.3251 | 24.0 | 1200 | 1.5916 | 0.9667 |
| 0.2744 | 30.0 | 1500 | 1.5880 | 0.9741 |
### Framework versions
- Transformers 4.32.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3
|